A double-blind placebo-controlled study to evaluate endometrial safety and gynaecological symptoms in women treated for up to 5 years with tamoxifen or placebo – A substudy for IBIS I Breast Cancer Prevention Trial ^☆

• Tiina Palva^a, , , 
• Hannu Ranta^a, b, 
• Anna-Maija Koivisto^c, 
• Liisa Pylkkänen^a, d, 
• Jack Cuzick^e,
• Kaija Holli^f

• ^a Pirkanmaa Cancer Society, Tampere, Finland
• ^b Pihlajalinna Ltd., Tampere, Finland
• ^c School of Health Sciences, University of Tampere, Tampere, Finland
• ^d Department of Oncology, University of Turku, Turku, Finland
• ^e Centre for Epidemiology, Mathematics and Statistics, Cancer Research UK, London, UK
• ^f University of Tampere, Tampere, Finland

• http://dx.doi.org/10.1016/j.ejca.2012.06.015, How to Cite or Link Using DOI

• Permissions & Reprints

---

Keywords

• Gynaecological symptoms; 
• Endometrial safety; 
• Tamoxifen; 
• Breast cancer prevention

---

1. Introduction

The finding of a decrease in the contralateral breast cancer incidence following tamoxifen administration for adjuvant therapy has led to its use in breast cancer prevention. Four large prevention trials with tamoxifen have been published. Initially, in the National Surgical Adjuvant Breast and Bowel Project’s (NSABP) Breast Cancer Prevention Trial, tamoxifen reduced the risk of invasive breast cancer by 49%.¹ Three other randomised prevention trials comparing tamoxifen with placebo have shown reduction of 38% in the overall breast cancer incidence and reduction of 48% in the oestrogen receptor-positive breast cancer.²

One important determinant of tamoxifen safety is the endometrium. The first report connecting tamoxifen and endometrial cancer was published in 1989³ and since then several reports have described the endometrial changes in postmenopausal breast cancer patients treated with tamoxifen.^{4, 5 and 6} The most common endometrial changes include endometrial polyps (with the incidence between 8–36% versus 0–10% in untreated women and endometrial hyperplasia (1.3–20% versus 0–10%). The risk for the endometrial carcinoma was 1.3–7.5-fold risk compared to untreated women. Also cystic atrophy/adenomyosis and growth of leiomyomas have been reported.⁷ At present, detailed guidelines give recommendations for the follow-up of patients taking tamoxifen for the treatment of breast cancer.⁸ Currently no active screening for asymptomatic patients treated with tamoxifen, other than routine annual gynaecological surveillance, is recommended.⁷

In the prevention studies with tamoxifen, the risk for endometrial cancer has been estimated to increase by 2.5-fold relative to placebo.² Further, the findings from the NSABP Breast Cancer Prevention Trial showed that the risk was increased only in postmenopausal women, but no increase in premenopausal women was observed. It has also been postulated, that even though the frequency of adverse events is in general reduced after stopping the tamoxifen treatment, the adverse effects on endometrium may persist.⁸ There are also recommendations for the follow-up of endometrial safety when tamoxifen is used in the preventive setting.⁹ Since uterine safety becomes more important when only a small benefit of the treatment is to be expected, as in the use of tamoxifen for breast cancer prevention, more active surveillance is needed, including annual vaginal ultrasonography.⁷ The more recent United States (US) guidelines for subjects taking tamoxifen for preventive setting include a baseline gynaecological examination before starting tamoxifen, and the annual follow-up thereafter, and also emphasise continuing surveillance post-treatment. Since tamoxifen-induced endometrial changes only appear after 2–3 years from the start of the treatment, annual transvaginal sonography may be started after that⁷ However, routine endometrial biopsy is not needed in the absence of abnormal vaginal bleeding.^7 and 10

In the first International Breast Cancer Intervention Study (IBIS I), a total of 7145 women at risk of breast cancer were randomised to receive either tamoxifen 20 mg or placebo for 5 years.¹¹ After a median follow-up of 96 months after randomisation, the risk ratio for breast cancer was 0.73 (95% confidence intervals (CI) = 0.58–0.91; p = 0.004) for those treated with tamoxifen compared to those treated with placebo.¹² The results also indicated that even though the risk-reducing effect of tamoxifen appeared to persist for at least 10 years, the majority of the adverse effects did not continue after the 5-year treatment period. A total of 28 endometrial cancers were reported, 17 in the tamoxifen group and 11 in the placebo group (risk rates (RR) 1.55; 95% CI = 0.68–3.65). Twelve of the cancers in the tamoxifen group but only three in the placebo group were detected during the active treatment period. Detailed data on other endometrial safety findings were not reported.

Even though the adverse effects of tamoxifen, including uterine cancer and endometrial hyperplasia, are well documented, there are only few data available from clinical studies on endometrial findings with tamoxifen use in preventive setting. This led us to perform this prospective, randomised, double-blind gynaecological sub-study to obtain detailed data on gynaecological symptoms and signs and to evaluate long-term endometrial effects.

2. Materials and methods

2.1. Subjects

This study was a sub-study for the IBIS I-study, in which in total 7145 women aged 35–70 years at the risk of breast cancer were randomised to receive either tamoxifen (TAM) 20 mg or placebo (PLA) for 5 years in randomised double-blind fashion. In Finland, altogether 136 women were included in the main study. Of these women 100 were eligible in this endometrium safety study (i.e. had intact uterus), and eventually 96 participated; two refused and the reason for non-participation was unknown for two women. The subjects were randomised between 15th February 1995 and 2nd November 2000. Prior to any study related procedures, the subjects gave their written consent. The study protocol was approved by the Pirkanmaa Hospital District Ethics Committee (24th January 1994, No. 94011M).

Of the 96 included women, 45 were treated with tamoxifen and 51 with a corresponding placebo. The subjects were followed-up clinically from randomisation up to 6 years, or premature discontinuation due to withdrawal of consent, breast cancer or other reason, such as hysterectomy. For occurrence of gynaecological malignancies the subjects were followed-up to at least 9 years (9–14 years).

2.2. Methods

The gynaecological follow-up with trans-vaginal ultrasound was performed by an experienced gynaecologist at baseline, at 2.5 years, at 5 years (end of treatment) and at 6 years. The endometrial thickness was measured as double-layer including potentially abnormal tissues at the widest point across the cavity between endometrial-myometrial interfaces. Endometrial biopsies were performed at baseline, at 2.5 and 5 years. Throughout the study all serious adverse events, gynaecological complaints and referrals to the hospitals were recorded. Extra gynaecological follow-up visits were arranged, if needed.

The information on gynaecological cancers diagnosed in the study subjects between 1st February 1995 and 21st July 2009 were retrieved from the Finnish Cancer Registry (FCR) database, by linking the study database with the FCR database. FCR is a population-based cancer registry with practically complete coverage of cancers diagnosed in Finland.¹³

2.3. Statistical analyses

Data are described using mean and standard deviation for normally distributed quantitative variables and median with minimum and maximum value for skewly distributed quantitative variables. For categorical variables, number of observations and percentages are reported. Because of relatively small sample size, differences between groups were tested only for selected variables using Mann–Whitney test (quantitative skewly distributed variables) or χ²-test/Fisher’s test (categorical variables). Rates for extra gynaecological events (extra visits, referrals to hospital, fractional curettage or hysteroscopic resection) were calculated by dividing the number of observed events by the number of cumulative women-years (WYs) separately for each group to be compared. Risk rates (RR) were calculated as the rates of the groups to be compared and 95% confidence intervals (95% CI) are given for risk rates. Calculations were performed with Statistical Package for the Social Sciences (SPSS) for Windows version 17 statistical software and with Stata 8.2 for Windows (RRs and confidence intervals).

3. Results

3.1. Baseline characteristics

For all 96 included study subjects, the groups did not differ, including the postmenopausal status or the use of hormonal replacement therapy (HRT) (Table 1). At baseline, majority of the subjects did not have any postmenopausal symptoms. The main abnormality in the clinical gynaecological status was the occurrence of uterine myomas; 18% in the TAM group and 12% in the PLA group (Table 2). The endometrial thickness (evaluable only in postmenopausal subjects) was similar between the groups (Table 2). 47% in the TAM group and 31% in the PLA group had abnormal findings in gynaecological ultrasound, mainly findings suggestive for uterine myomas. Due to baseline findings altogether 7 (16%) subjects in the TAM group and 3 (6%) subjects in PLA group were referred to the hospital, mainly for gynaecological fractional curettages (Table 2). At baseline endometrial curettage was performed for two postmenopausal women, both in the TAM group.

3.2. Study discontinuations and follow-up years

Women in the TAM group discontinued the study significantly more often compared with those in the PLA group (44% versus 22%, p = 0.017). Altogether 40 subjects in the PLA group and 24 subjects in the TAM group remained in the study for up to 6 years. The discontinuations occurred significantly earlier, with the median of 15 months in the TAM group compared with the median of 30 months in the PLA group (p = 0.044). For both groups, the discontinuation rates and the times of discontinuations were independent of the menopausal status (Table 3). In the TAM group, there were discontinuations during the whole treatment period, but particularly during the first year, unlike in the PLA group (Fig. 1). The main reasons for discontinuation in the TAM group were vasomotor symptoms for 10 and 8 subjects had more than one reason for discontinuation. In the PLA group no one discontinued due to vasomotor symptoms and everyone had only one reason for discontinuation (Table 3). During the study there were no major differences in the use of HRT between the groups. However, the use of vaginal oestrogens was more common in the TAM group (data not shown). No differences between the groups could be seen in number of breast cancer or hysterectomies (Table 3).

There were in total 183 cumulative WYs in the TAM group and 266 in the PLA group. For the premenopausal women there were 124 WYs in the TAM group and 157 in the PLA group, and for the postmenopausal women the corresponding figures were 59 WYs and 109 WYs, respectively.

3.3. Gynaecological findings

In the clinical gynaecological status there were only a few abnormal findings other than suspected myomas during the follow-up. The median endometrial thickness for postmenopausal women at 5 years was 4.3 mm (2–9) in the TAM group and 2.0 mm (1–4) in the PLA group, respectively, with statistically significant difference between the groups (p = 0.011). However, one year after the end of treatment, the endometrial thickness was equal between the groups (Table 4). At 5 years, the endometrial thickness was less than 5 mm for 63% (n = 5) in the TAM arm, whereas in the PLA arm it was less than 5 mm for all 15 women (p = 0.032). During the routine endometrial biopsy, the endometrial or cervical polyp was diagnosed in 6 subjects in the TAM group and in 3 subjects in the PLA group.

3.4. Extra gynaecological events

During the treatment period, the number of extra gynaecological visits and the number of hospital referrals per patient were significantly higher in the TAM group (Table 5). Endometrial curettages were also significantly more common in the TAM group (RR = 3.49, 95% CI 1.14–12.64), and the difference in the curettage rate between the groups was even more clear for premenopausal women (RR = 4.22, 95% CI 1.09–23.86). Endometrial curettages were performed for two postmenopausal women in the PLA group and two women in the TAM group, all referred at 2.5 years. The number of hysterectomies during the follow-up was 1 in TAM group and 4 in PLA group (Table 5). All but one reason for the hysterectomy were myomas.

3.5. Gynaecological malignancies

When the study files were linked with the Finnish Cancer Registry database, three gynaecological malignancies were diagnosed between 1st February 1995 and 21st July 2009, all in the TAM group. Two of these subjects had discontinued the study medication at 8 months (endometrial cancer) and at 10 months (ovarian cancer) due to hot flushes. One further subject was diagnosed with endometrial carcinosarcoma 120 months after the start of TAM treatment, which was continued for up to 60 months. She had received referral for endometrial curettage at 5 years follow-up visit and was diagnosed with endometrial polyp.

4. Discussion

The randomised, double-blind, placebo-controlled design of this study enabled us to investigate the endometrial safety in an unbiased fashion. Even though the study is small, it is well characterised and includes detailed and comprehensive follow-up information for endometrial safety. Further, the patients were followed-up by the same oncologist and gynaecologist. 96% of the eligible IBIS I study subjects also participated in this gynaecological sub-study. To our knowledge, this kind of systematic histological evaluation of the endometrium has not been performed in earlier studies including subjects who have used tamoxifen in preventive setting, although there are several studies in adjuvant setting.^{3, 4, 5, 6 and 7}

Our result showed that women in the TAM group discontinued the study significantly more often than those in the PLA group, irrespective of their menopausal status. Even though offering endometrial surveillance has been suggested to improve compliance, and thus to be the way forward to continue long-term tamoxifen treatment for breast cancer prevention,⁷ this was not for our study – Only few discontinuations in our study were related to gynaecological findings, but were mainly attributable to the vasomotor symptoms. The discontinuation rate was high in the TAM group during the first year of treatment, while in the PLA groups none of the subjects discontinued within that time. In addition, most subjects in the TAM group had more than one reason for discontinuation reflecting the problems with tamoxifen use. Higher discontinuation rates in the tamoxifen group have been reported also in the main IBIS study, in which 64% of women in the tamoxifen group completed the full 5 years of treatment, whereas the corresponding figure was 72% in the placebo group.¹²

While analysing the potential risks and benefits of a particular treatment, we also need to keep in mind the compliance, which may be poor particularly in the preventive setting. A recent meta-analysis on breast cancer chemoprevention decisions showed that perceived vulnerability to breast cancer was consistently correlated with increased uptake of the treatment, and concern for adverse effects was correlated with reduced uptake. The authors concluded that breast cancer chemoprevention uptake rates are low and variation is wide.¹⁴ It is thus obvious, that only a minority of women at risk take the opportunity to reduce the risk for breast cancer, in those countries where tamoxifen is approved for breast cancer prevention.

The key issue for breast cancer chemoprevention is the risk–benefit ratio, and the key contributors to the risk are the serious adverse events, including increased risk for endometrial cancer. In the main IBIS I study after a median follow-up of 96 months after randomisation, 142 breast cancers were diagnosed in the 3579 women in the tamoxifen group and 195 in the 3575 women in the placebo group (4.97 versus 6.82 per 1000 women-years, respectively; RR = 0.73, 95% CI = 0.58–0.91; p = 0.004).¹¹ The prophylactic effect of tamoxifen was fairly constant for the entire follow-up period, and no diminution of benefit was observed for up to 10 years after randomisation.¹¹ Long-term breast cancer prevention by tamoxifen is also reported in the 20-year follow-up of the Royal Marsden prevention trial.¹⁵ Even though the most side effects of tamoxifen do not continue after the 5-year treatment period,¹¹ the possible endometrial cancer risk may still persist ^8 and 16and this increased risk is observed in particular in postmenopausal women. Probably due to the small sample size we could not see either the decreased breast cancer risk or the increased endometrial cancer risk.

The results of the current study showed that the median endometrial thickness in postmenopausal women was significantly increased at 5 years in the TAM group, but there was no difference between the groups within one year after discontinuation of the treatment. The endometrial thickness was significantly increased, even though the majority of those who discontinued in the TAM group, did so prior to the first planned trans-vaginal ultrasound follow-up at 2.5 years.

There were also significantly more referrals to hospitals due to endometrial symptoms and extra gynaecological visits and gynaecological events in the TAM group. During the treatment period, 27% of the women randomised to the TAM group had endometrial curettage, compared to 10% in the PLA group. Considering the women years, the curettage rate for premenopausal women in the TAM group was even 4.2-fold. Even though there were significantly more extra investigations and procedures in the TAM group, the results of these investigations were benign in nature. More specifically, no significant findings were observed in the endometrial biopsies. For example, the endometrial biopsies of those three women diagnosed with endometrial cancers, did not show any premalignant or otherwise suspicious changes prior to cancer diagnosis. The results support the current concept that for non-symptomatic women endometrial biopsies are not needed.

Chemoprevention of breast cancer by tamoxifen is particularly targeted for premenopausal women, where gynaecologic and thromboembolic side effects are in general less common compared to postmenopausal women.¹² In the current study, there were no severe side effects for premenopausal women, but, however, 40% of the premenopausal women in TAM group discontinued the study after median treatment time of only 15 months. Compared to placebo, they had significantly more gynaecological events and the curettage rate was as high as 4.2-fold. Endometrial curettages were performed for 10 of the 29 premenopausal women in the TAM group.

5. Conclusions

The discontinuation rate in the TAM group was twice as high as in the placebo group suggesting that about half the discontinuations are due to side effects. The discontinuations occurred early and were mainly due to vasomotor symptoms, reflecting the preventive nature of the study. Even though there were significantly more gynaecological events, and the endometrial thickness was significantly increased during the TAM treatment, no increase in serious adverse events could be observed. Our results do not support routine gynaecological surveillance in asymptomatic women even in preventive setting.

Conflict of interest statement

There are no conflicts of interest for any of the authors.