Correspondence: Dr F. Al-Azzawi, Department of Obstetrics and Gynaecology, University of Leicester, Robert Kilpatrick Clinical Sciences Building, Leicester Royal Infirmary, Leicester LE2 7LX, UK.
Objective To assess the value of identifying endometrial structural abnormalities at baseline hysteroscopy in predicting the pattern of bleeding in postmenopausal women treated with hormone replacement therapy.
Design A randomised, double-blind, dose-ranging study.
Setting A teaching hospital in the UK.
Population One hundred and seventy-six healthy postmenopausal women.
Methods Women were randomised to receive one of four doses of oral trimegestone (0.05, 0.1, 0.25 and 0.5 mg per day), from day 15–28, and a daily dose of 2 mg oral micronised oestradiol for six treatment cycles. Women completed diaries in which the bleeding episodes were recorded. Hysteroscopy under local anaesthesia and endometrial biopsy were performed at baseline and on day 24 of the last treatment cycle.
Results Women with submucous fibroids had more prolonged (P= 0.026) and heavier (P= 0.002) progestogen-associated bleeding (odds ratio 4.54). The incidence of intermenstrual bleeding, but not its duration or severity, was higher in women with submucous fibroids (P= 0.017). There was a clear dose-dependent effect of trimegestone, with a consistently later onset of progestogen-associated bleeding occurring with increasing doses of trimegestone (P < 0.001), and such episodes became progressively lighter and of shorter duration over time (P < 0.001).
Conclusion Hysteroscopic evaluation of the endometrial cavity in women treated with hormone replacement therapy, predicts the occurrence of heavy and unscheduled bleeding.
The use of hormone replacement therapy (HRT) in postmenopausal women has been limited by medication-induced adverse effects, and in particular, the induction of uterine bleeding1. In 50% of women uterine bleeding, particularly if unpredictable or heavy, is the most common cause of noncontinuation of HRT2. The long term use of HRT has far-reaching effects in improving the quality of life, preventing osteoporosis, cardiovascular disease and Alzheimer's disease3–5. Therefore, factors responsible for unscheduled bleeding, if identified and treated, may enhance the provision of a healthy existence in post-reproductive years.
Hysteroscopically identified endometrial structural abnormalities in postmenopausal women are associated with unscheduled uterine bleeding while on HRT. Of all fibroids, the submucous type seems to be the most symptomatic6. In our previous report 47.6% of women with irregular or heavy bleeding on HRT had structural lesions on hysteroscopic assessment7. Similar results (46.7%) were reported by Nagele et al.8. Akkad et al.9 found that the presence of submucous fibroids was associated with a three-fold increase in the risk of abnormal uterine bleeding in premenopausal women and a two-fold increase in the risk of abnormal withdrawal bleeding in peri- and postmenopausal women using HRT. While endometrial polyps are invoked as a cause of abnormal bleeding patterns10, our previous reports failed to show such a correlation7–9.
Trimegestone is a novel norpregnane progestin with potent progesterone receptor and very low androgen receptor affinities, and no detectable affinity to oestrogen receptors11. Such a profile makes this progestogen a suitable compound to investigate in combined HRT. We have demonstrated that the dose of trimegestone can dramatically influence the pattern of bleeding12. Nevertheless, some of the women who received the higher doses of trimegestone experienced early onset, heavy and prolonged progestogen-associated bleeding, while others who received the lower doses had late onset and light menstrual discharge of short duration, and therefore, mechanisms other than the dose of trimegestone seem to be relevant in modulating endometrial shedding.
The aim of this study was to assess the predictive value of hysteroscopically diagnosed endometrial structural abnormalities on the pattern of bleeding in post-menopausal women treated with trimegestone-based sequential HRT.
Women recruited into this study were referred by their general practitioners to the Menopause Research Unit, Leicester University School of Medicine and Bioscience. The protocol was approved by the local ethics committee and all women signed their informed consent. This cohort of 176 women, was part of a multicentre double-blind dose-ranging study population of 256 women.
Women included in this study were healthy, aged 45–65 years, with an intact uterus, whose menopausal status required to fulfil one of the following criteria:
1At least six months had elapsed since the last menstrual period, with follicle-stimulating hormone and oestradiol in the postmenopausal range.
2Had used HRT continuously for more than two years.
3Had been on HRT for at least one year with pre-treatment follicle-stimulating hormone and oestradiol levels in the postmenopausal range.
None had received any form of sex steroid treatment for six weeks before the commencement of study medications. Those who had ever used oestradiol implants were excluded. Tests for liver and renal function were performed and those women with abnormalities were excluded. All women over the age of 50 had a normal mammogram within three years, and normal cervical smears within the previous six months. General, breast, and pelvic examinations were conducted to confirm normality. Pelvic sonography was performed to measure the size of the uterus, endometrial thickness and to assess the adnexa. All endometrial biopsies were evaluated by two independent pathologists, blinded to the dose of trimegestone.
To evaluate the presence of endometrial structural abnormalities, all women underwent diagnostic hysteroscopy under local anaesthesia, as previously described7. Submucous fibroids and endometrial polyps were differentiated by their appearance and consistency. Submucous fibroids were graded according to the degree of protrusion into the endometrial cavity into 3 grades; I, II and III, depending on how much of an estimated sphere is protruding into the endometrial cavity: up to 1/3, up to 2/3 or > 2/3, respectively. The endometrium was sampled using the Vabra curette and women with evidence of hyperplasia or carcinoma were excluded. Additional sex steroid treatments were not allowed during the study and women who were using treatments known to interfere with steroid metabolism were excluded from the study. At the final visit, general, breast, and pelvic examination were repeated, pelvic sonography, hysteroscopy under local anaesthesia and endometrial biopsy were performed on day 24 of the last treatment cycle.
In the dose-ranging, double-blind, parallel groups study, women were randomised to receive treatment medications for six cycles. The study treatment was randomly assigned to the randomisation numbers (using SAS statistical package, SAS institute Inc, Cary, North Carolina, USA). The four doses of trimegestone were randomised on a 1:1:1:1 basis. Women were numbered consecutively in the order in which they entered the study.
Each treatment cycle consisted of oral micronised oestradiol 2 mg/day (RU 3499) for 28 days and oral trimegestone (Hoechest Marion Roussel, Romianville, France) (0.05, 0.1, 0.25, or 0.5 mg) from days 15.28. The first visit was a pre-study assessment for inclusion, 2–3 weeks prior to commencement of 168 days treatment. Women attended the unit at the end of the third cycle of treatment and on the 24th day of the sixth treatment cycle. If a woman could not attend on that day, a seventh treatment cycle was prescribed.
Women completed menstrual diaries for each treatment cycle, in which the severity of the bleeding was subjectively scored as: 0 = no bleeding, 1 = spotting, 2 = slight bleeding, 3 = moderate bleeding, and 4 = heavy bleeding. The total bleeding score was taken as the sum of the daily scores for each cycle. The diaries were collected at each visit and were checked for accuracy and completeness. These diaries also served as an additional parameter to check compliance since a box had to be ticked when the tablets were taken. The medication packs were collected and any remaining tablets were counted. Where less than 80% of the medications were used, noncompliance was marked and the woman withdrawn from the study. All adverse events and concomitant medications were checked at each visit.
A bleeding episode was defined as an episode of uterine bleeding which lasts for one or more days with at least one clear day before and after. The progestogen-associated bleeding was defined as the longest bleeding episode which starts between day 22 of one cycle and day 7 of the next treatment cycle inclusive. All other bleeding episodes were called intermenstrual bleeding.
Only completed diaries were analysed for the purpose of this study. The bleeding data in cycles in which the endometrial biopsy was taken were excluded. The day of commencement of the oestrogen phase of the cycle was deemed as a fixed point of reference (i.e. day 1).
Data were collected on up to seven treatment cycles and were analysed using mixed effects regression models. For continuous data, such as day of onset of progestogen-associated bleeding, duration and severity of the bleeding, the residual maximum likelihood was used13 to fit the model. For binary data, such as the occurrence of intermenstrual bleeding, a generalised linear mixed model with binomial errors was used14. Calculations were made using GENSTAT15.
Between November 1994 and December 1995, 176 women were randomised to one of the four dose groups of trimegestone, 169 were treated and 131 completed the study, while only 129 women were evaluated for endometrial analysis, as two women declined to have a biopsy at the end of the study. There were seven women who did not start treatment after randomisation and were withdrawn from the study as a result of being; lost to follow up (n= 3), protocol violation (n= 3) and adverse events (n= 1).
Women randomised into the study had an overall mean (SD) age of 52.5 years (5.1), weight of 64.2 kg (10.5), height of 160.9 cm (6.7) and body mass index of 24.8 kg/cm2 (3.4) (Table 1). There were no differences between the treatment groups in terms of age, race, height, weight, body mass index and smoking habits. The number of women who completed the study were: 33, 33, 37 and 28 in dose groups 0.05 mg, 0.1 mg, 0.25 mg, and 0.5 mg, respectively. Thirty-eight women did not complete the study due to: bleeding problems (n= 9), mastalgia (n= 6), gastrointestinal tract symptoms (n= 1), headaches (n= 4), depression (n= 2), pruritus (n= 3), muscular cramps (n= 1), weight gain (n= 1), weight loss (n= 1), gastrointestinal tract cancer (n= 1), breast cancer (n= 1), deep venous thrombosis (n= 2) women's request (n= 4) and lost to follow up (n= 2). There was no statistically significant difference between the number of women who withdrew from each trimegestone dose group.
Table 1. Demographic characteristics of the treated women, and the number of withdrawals in each dose group. Values are given as n or mean (SD). BMI = body mass index.
No. of women treated
Given up smoking
Never a smoker
Women who completed
Six hundred and ten diaries were analysed. All women who completed the study experienced an episode of bleeding in each treatment cycle. The diaries were evaluated for the day of onset, duration and severity of the progestogen-associated bleeding and inter-menstrual bleeding episodes. Submucous fibroids were identified at baseline hysteroscopy in 4, 7, 5 and 7 women in the 0.05 mg, 0.1 mg, 0.25 mg and 0.5 mg dose groups, respectively (Fig. 1). Endometrial polyps were identified in 3, 3, 5 and 2 women assigned to 0.05 mg, 0.1 mg, 0.25 mg and 0.5 mg dose groups, respectively. Age, duration of the menopause, previous use of HRT, height, weight, or body mass index did not influence the bleeding pattern among the four trimegestone dose groups.
Analysis of the bleeding diaries showed that the episodes of progestogen-associated bleeding lasted longer (P= 0.026) and were heavier (P= 0.002) in the presence of submucous fibroids (Table 2). This effect was apparent only in the higher three doses of trimegestone. The odds of having mean total bleeding score of over 17.75 (the 3rd quartile) was more than four times higher in women with submucous fibroids than without (odds ratio 4.54, 95% CI 1.45, 14.26).
Table 2. The day of onset, duration, and severity, of the progestogen-associated bleeding patterns in women with and without submucous
Dose groups (mg)
No. of women
Day of onset
Women with submucous fibroids
Women without submucous fibroids
Intermenstrual bleeding occurred more frequently in the presence of submucous fibroids (P= 0.017) (Fig. 2). However, there was no statistically significant association between submucous fibroids and the duration (P= 0.83) or the severity (P= 0.27) of the intermenstrual bleeding (data not shown). There was no statistically significant effect of the endometrial polyps on any of the bleeding parameters studied.
The progestogen-associated bleeding episodes occurred progressively later in the treatment cycle (P < 0.001) with the increase in the dose of trimegestone. The progestogen-associated bleeding episodes were consistently shorter (P < 0.001) and lighter (P < 0.001) in the higher doses (Table 1). The episodes of intermenstrual bleeding were lighter in higher doses (P= 0.035), but there was no demonstrable dose-response effect on the incidence or the duration of the intermenstrual bleeding.
The histological assessment of the endometrium at the end of the study showed secretory changes (96%), proliferative changes (1.6%), inactive endometrium (0.8%), inadequate samples for histological assessment (0.8%), and one case of simple hyperplasia which was found in a woman receiving 0.5 mg of trimegestone. There was no statistically significant difference in the histological type of the endometrium between the four treatment groups.
The pathogenesis of abnormal uterine bleeding in association with submucous fibroids is not fully understood and a number of explanations have been suggested. Sampson (1912)16 found that there were structural and branching abnormalities in the arterial blood supply to these fibroids, although an obstruction to the uterine venous plexus can lead to generalised or focal endometrial venous ectasia and consequently more prolonged and heavier blood loss17. This however remains an unconfirmed possibility.
This study demonstrates that hysteroscopically identified submucous fibroids are predictive of heavier and longer progestogen-associated bleeding, with a higher incidence of intermenstrual bleeding. The effect of submucous fibroids on the duration and severity of the progestogen-associated bleeding episodes was not apparent in the lowest dose of trimegestone (0.05 mg), which may be due to fewer women with submucous fibroids in this group. Alternatively, there may be a dose threshold below which the dose-response relationship no longer exists.
In the generalised linear mixed model, the occurrence of submucous fibroids in all women in this cohort was considered together, and therefore allowed for the identification of submucous fibroids as a significant predictor of the heavy, prolonged progestogen-associated bleeding and of intermenstrual bleeding. However, the number of women with submucous fibroids in each dose group was too small to allow a separate analysis of the effect of the number and grade on the pattern of bleeding. The lack of association of the unscheduled pattern of bleeding with endometrial polyps confirms our previous reports7,9.
The dose of trimegestone did not affect the incidence of intermenstrual bleeding, while the presence of submucous fibroids predicted its occurrence. Therefore, we advocate hysteroscopic assessment rather than changing the dose of trimegestone in women with intermenstrual bleeding on hormone replacement therapy.
While it may be difficult to argue the necessity of performing hysteroscopy for all women contemplating HRT, it is necessary to evaluate the endometrial cavity hysteroscopically to exclude the presence of submucous fibroids in those who experience heavy or unscheduled bleeding while receiving HRT. Removal of submucous fibroids by local excision such as laser ablation or electro-resection may rectify the abnormal bleeding pattern in these women18.
In our previous report of the whole study population of 256 women12, randomised to the four trimegestone groups, a dose-dependent modulation of uterine bleeding was evident. Higher trimegestone doses induced later day of onset, progressively lighter and shorter episodes of progestogen-associated bleeding. The analysis of the bleeding diaries in this study of 176 women treated in our centre, confirmed the dominant effect of the dose of trimegestone on the day of onset, severity and duration of the progestogen-associated bleeding.
The incidence of simple cystic hyperplasia in women on hormone replacement therapy is 7%19, and there was one case of simple hyperplasia in the 0.5 mg dose, which could have risen by chance. In this study, it was not possible to evaluate a potential interaction between the dose of the trimegestone and the presence of submucous fibroids in view of the small number of women with submucous fibroids in each dose groups.
In conclusion, we recommend hysteroscopic assessment of women with abnormal bleeding patterns on HRT regimen, to exclude the presence of submucous fibroids.
This study was supported by Hoechest Marion Roussel Romainville, France and Wyeth-Ayerst, USA, in a joint development programme.