Endometrial histomorphometry of trimegestone-based sequential hormone replacement therapy: a weighted comparison with the endometrium of the natural cycle

May Wahab1, John Thompson1, Bushra Hamid2, Sue Deen3 and Farook Al-Azzawi1,4

1 Gynaecology Research Group, Department of Obstetrics and Gynaecology, Robert Kilpatrick Clinical Sciences Building, Leicester Royal Infirmary, PO Box 65, Leicester LE2 7LX, 2 Pathology Department, Trafford General Hospital, Manchester M41 5SL and 3 Pathology Department, West Suffolk Hospital,Bury St Edmunds IP33 2QZ, UK


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Histomorphometric changes in the endometrium were evaluated under the effect of a trimegestone-based sequential hormone replacement therapy (HRT) regimen, and the findings were compared to those in endometrium of the natural cycle. Endometrial samples were taken from postmenopausal women who completed a randomized, double blind, dose-ranging study of oral trimegestone (0.05, 0.1, 0.25 and 0.5 mg per day) from day 15 to day 28 with continuous micronized oestradiol 2 mg daily for six treatment cycles. The HRT-treated endometrium, irrespective of the dose, had a smaller mean total glandular area, smaller average glandular diameter, smaller mean total vascular space area and diameter than the luteal phase. Stromal cellularity was similar in the four dose groups. There were reduced glandular secretions in the endometrium from the high dose group. The relative weighting of these histological parameters was evaluated by linear discriminant analysis. The weighted values were dose independent, and may therefore represent either a specific effect of trimegestone, number of days administered, or both. We have constructed an equation to assign a value for a histological parameter which determines the position on linear discriminant functions. These assigned values can be used in other sequential HRT regimens to determine the relative influence of a given progestogen on endometrial morphology in relation to different phases of the natural cycle.

Key words: endometrial histomorphometry/hormone replacement therapy/linear discriminant analysis/postmenopause/trimegestone


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
The endometrium of the natural cycle is associated with morphological changes and ends, in the absence of implantation, in endometrial shedding and menstruation. Several authors have tried to define these endometrial changes as early, mid and late stages of the proliferative or secretory development. However, the histological definition of the phase of endometrial development is based on the whole picture and not on one single parameter since in most tissue sections a greater or lesser extent of mixed structural changes occur. Therefore, when ascribing a particular phase of endometrial development, histopathologists label the tissue, in the presence of mixed structural changes, according to the most advanced feature. Further, regional variations in endometrial responses to sex steroids are recognized such that endometrial tissues from the uterine fundus give more reliable information than the isthmus, which may add a further complexity in the histological analysis of endometrial curetting samples. Moreover, the depth of the endometrial sample examined is equally critical, for example, biopsies that include only the basalis but not the functionalis are not datable (Noyes et al., 1950Go).

Exogenous sex steroids result in withdrawal bleeding and induce changes in all components of the endometrium, which are different from those under the influence of the natural ovarian cycle (Johannisson et al., 1982Go; Johannisson, 1990Go; Li et al., 1992Go; Casanas-Roux et al., 1996Go; Dallenbach-Hellweg and Poulsen, 1996Go; Habiba et al., 1998Go). The effects of contraceptive steroids on the endometrium depend on the type, dose and duration of the steroid administered. Norethisterone, 50 µg/day administered as a contraceptive through a vaginal ring, does not affect the size or number of the glands per unit area, but when the dose is increased to 200 µg/day, it significantly reduces the size of the individual gland at 6 weeks, and reduces the number of these glands if given for 10 weeks (Johannisson, 1990Go).

Most studies of endometrial morphology suffer from a small sample size, short duration of treatment or an unclear method of assessment. The effect of levonorgestrel (LNG) administered through a vaginal ring as a contraceptive for 90 days has been studied (Johannisson et al., 1991Go). LNG significantly reduced the glandular diameter, glandular volume density and the number of the arteries in the endometrial tissue. However, it is not clear how the last conclusion was drawn since no mention of the method of assessment of the endometrial vascularity was made. In another study the morphological changes were examined with different progestogens and for different duration of treatment, but there was only one patient in each group (Ludwig et al., 1982).

There is paucity in the literature regarding the histological changes associated with HRT in postmenopausal women. The morphometric characteristic associated with oral oestrogen and cyclically administered progesterone as vaginal gel has been studied and an increase in glandular coiling in response to this regimen has been found which mimicked the glandular changes in the luteal phase (Casanas et al., 1996). The histological changes associated with cyclical sequential combined orally administered oestrogen and norethisterone were also assessed (Habiba et al., 1998Go). Nevertheless, these studies did not address the question of the contribution of individual morphological findings to the overall assessment of endometrial development in the HRT-treated women compared to the endometrium of the natural cycle.

Trimegestone is a novel norpregnane progestin, which in human recombinant receptor binding studies demonstrates high affinity to progesterone receptor, very low affinity to androgen receptor, and no detectable affinity to oestrogen receptor (Bouchoux, 1995Go). Trimegestone is being developed for use in conjunction with oestrogen in HRT. In a recent study (Al-Azzawi et al., 1999Go), we have demonstrated that there are dose-dependant changes in the bleeding pattern in women treated with four doses of sequentially administered trimegestone (0.05, 0.1, 0.25 and 0.5 mg/day) for 6 months; however, routine histological assessment of the endometrial biopsies at the end of the study showed a secretory picture in 96% of women. Therefore, such histological assessment does not help to explain the different bleeding patterns.

In this study we have evaluated detailed histological features relevant to endometrial developments in women treated with trimegestone-based HRT and during the natural cycle using image analysis techniques and linear discriminant analysis to examine: (i) specific morphological changes induced in the endometrium by the sequentially administered trimegestone in a dose-ranging study; (ii) to compare the findings to those measured in five definable stages of endometrial development of the natural cycle; (iii) the potential existence of specific structural changes in the endometrium under this trimegestone-based HRT regimen which may be associated with an observed bleeding pattern.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
In a randomized, double-blind, dose-ranging study, postmenopausal women were given oral trimegestone (0.05, 0.1, 0.25 and 0.5 mg per day) from day 15 day to 28 with continuous micronized oestradiol 2 mg daily for six treatment cycles. The protocol was approved by the local ethics committee and all patients signed an informed consent. This cohort of 176 women recruited in our centre is part of a multicentre, double-blind, dose-ranging study population of 256 women.

The inclusion criteria for women who volunteered to participate in this study were: healthy, aged 45–65 years with intact uterus who were at least 6 months postmenopausal; with follicle stimulating hormone (FSH) and oestradiol in the postmenopausal range; had had HRT for more than 2 years, or had been on HRT for at least 1 year with pre-treatment FSH and oestradiol concentrations in the postmenopausal range. None had received any form of sex steroid treatment for 6 weeks before the commencement of study medications. Those who had ever used oestradiol implants were excluded.

Endometrial biopsies were obtained on day 24 of the last treatment cycle. The endometrial samples were fixed immediately in 10% formal-saline embedded in paraffin and 4 µm sections were stained with haematoxylin and eosin for histological assessment (standard protocol at the Histopathology Department, Leicester Royal Infirmary).

The control samples were deep endometrial biopsies obtained from healthy, regularly menstruating women, aged 27–50 years (mean age ± SD, 38 ± 6.1), undergoing laparoscopic sterilization using sharp curette or from hysterectomy specimens none of which was indicated for a menstrual disorder. None of these women had received any hormonal treatments for 2 months prior to the procurement of the specimens (n = 37). All women were given urinary luteinizing hormone (LH) surge detection kit tests (First Response; Carter Wallace Ltd, Folkstone, UK), which were used during the month preceding the endometrial biopsy or hysterectomy. The histological diagnoses, were proliferative (n = 8), early luteal (n = 8), mid luteal (n =5), late luteal (n = 8) and menstrual (n = 8). These biopsies were dated both by LH surge and the date of the last menstrual period, and were examined by two independent pathologists who were blinded to the LH surge and menstrual dates. Where all agreed, the specimen was included as a control sample.

Assessment of the endometrial sections
Fifteen randomly selected fields (Hamilton, 1995Go) per slide (x200) were captured using an image analysis program to assess 13 variables. Glands: total glandular area, average glandular diameter, number of complete and incomplete glands per field, glandular epithelial height in the glands (columnar or cuboidal), regularity of the glandular epithelial surface (smooth or irregular), percentage of gland containing luminal secretions and subnuclear vacuoles, and number of invaginations (telescoping) in the field. Vascularity was examined for total vascular space area, average vascular space diameter, and number of vascular spaces per field. To assess stromal cellular density, the nuclei were counted in 15 randomly selected fields (Hamilton, 1995Go) per slide (x1000) under oil immersion. All fields examined were restricted to the functionalis layer.

Images were captured using Axioplan microscope (Carl Zeiss, Welwyn, Herts, UK), and a colour video camera (Sony CCD/RGB). The data presented are per unit area and the corresponding areas of an image captured with x200 and x1000 were 0.121 and 0.0046 mm2 respectively using a measurement graticule. The areas measured or cells counted were evaluated using the KS300 image analysis programme (Kontron Imaging Systems, Thame, Bucks, UK).

Statistics
For each specimen 15 fields were taken, and the measurements for each field were averaged to give a mean score and standard deviation for that specimen. Kruskal–Wallis non-parametric one-way analysis of variance (ANOVA) and Wilcoxon's rank sum test (Mann–Whitney test ) were used to compare individual measurements between the four dose groups and between phases of the natural cycle. Mann–Whitney test was used to compare the histological parameters between women who bled on the day of the biopsy and those who had not bled by then. The average scores for each specimen on 13 variables were then analysed by linear discriminant analysis (Mardia et al., 1979Go) to look for differences between the four treated groups and the five phases of the natural cycle.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
In our centre, 176 women (mean age ± SD, 52.5 ± 5.1 years) were randomized to one of the four dose groups and 131 completed the study; seven women who did not start treatment after randomization and were withdrawn from the study: lost to follow up (n = 3), protocol violation (n = 3) and adverse events (n = 1). Thirty-eight women did not complete the study, of whom only nine withdrew due to irregular bleeding. There was no statistically significant difference between the number of patients who were assigned or withdrew from each trimegestone dose group. 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. There was a dose-dependent effect on the pattern of bleeding as women on the higher doses of trimegestone had later mean day of onset of the progestogen-associated bleeding (PAB), which was lighter and shorter (Al-Azzawi et al., 1999Go). Two women declined to have a biopsy at the end of the study, and therefore the total number of endometrial samples obtained at the end of the study was 129. Endometrial biopsies of women who bled before the day of the biopsy were excluded (n = 12), and 24 other samples were too scanty for meaningful analysis. Ninety-three endometrial biopsies were evaluable for morphometric assessment (n = 20, 20, 28 and 25 in the 0.05, 0.1, 0.25 and 0.5 mg trimegestone dose groups respectively).

The endometrium of the natural cycle (Figure 1A–EGo)
The mean total glandular area (Figure 2aGo) was higher in the luteal phases compared to the proliferative phase (early luteal, P = 0.0002; mid luteal, P = 0.0007; late luteal, P = 0.003). The average glandular diameter (Figure 2bGo), was higher in the mid luteal and menstrual phases compared to the proliferative phase, but these differences were not statistically significant. The height of the glandular surface epithelium was lowest in the mid luteal (P < 0.004), but started to increase gradually in the late luteal and menstrual phases. The average number of the endometrial glands per unit area did not differ significantly in the five phases of the natural cycle (Figure 2cGo). Endometrial glandular epithelial vacuoles did show variation across the phases (P = 0.0002), and were significantly more numerous in the early luteal phase compared to any one of the other phases (all P < 0.004). There was some evidence of a difference in the glandular secretions between the phases of the natural cycle (P = 0.03; Figure 2dGo) due to slightly higher values in the mid and late luteal phases.



View larger version (0K):
[in this window]
[in a new window]
 
Figure 1. Endometrial sections of the five phases of the natural cycle, stained with haematoxylin and eosin, showing: (A) the proliferative phase, (B) early luteal phase, (C) mid luteal phase, (D) late luteal phase, and (E) menstrual phase. Scale bar = 70 µm.

 


View larger version (27K):
[in this window]
[in a new window]
 
Figure 2. (ah) Mean and standard deviation of areas and cell counts examined in endometrial samples obtained from the four dose groups of trimegestone ({alpha}, ß, {gamma}, {delta} corresponding to 0.05, 0.1, 0.25 and 0.5 mg respectively, filled circles), and the endometrium from the five phases of the natural cycle (1, 2, 3, 4, 5, corresponding to proliferative, early luteal, mid luteal, late luteal and menstrual, respectively, empty circles). (a) Total glandular area, (b) average glandular diameter, (c) number of complete glands, (d) percentage of glands with secretions, (e) total vascular space area, (f) average vascular space diameter, (g) number of vascular space area, and (h) stromal cellularity.

 
The mean total vascular space area increased gradually as the luteal phase advanced and was greatest in the menstrual phase (P < 0.05; Figure 2eGo). The average vascular space diameter varied between the phases of the natural cycle (Figure 2fGo). There was no suggestion of a difference in the mean number of vascular spaces per unit area in the five phases of the natural cycle (Figure 2gGo). There was no difference in the stromal cellularity between the different phases of the natural cycle (Figure 2hGo).

The trimegestone-treated endometrium
The endometrium of the HRT cycle, irrespective of the dose, had significantly smaller mean total glandular area compared to the luteal phase (P < 0.0001; Figure 2aGo), and slightly smaller average glandular diameter than the luteal phase (Figure 2bGo). There was no difference in the mean number of glands per unit area when compared with the natural cycle (Figure 2cGo). The height of the glandular epithelium was significantly lower compared to the phases of the natural cycle (P < 0.0004). Glandular telescoping (Figure 3AGo), was more prevalent in the 0.1 and 0.25 mg doses of trimegestone-treated endometrium, but was generally lower than the early and late luteal phases; however, it was difficult to draw any conclusion, since there was no telescoping in the proliferative or mid luteal phases. The higher dose of trimegestone was associated with significantly lower glandular secretions compared to the lower doses, or to the phases of the natural cycle (P < 0.005, Figure 2dGo and P < 0.04, Figure 3BGo respectively).



View larger version (0K):
[in this window]
[in a new window]
 
Figure 3. Endometrial sections of the trimegestone-based hormone replacement therapy, stained with haematoxylin and eosin, showing: (A) higher percentage of glandular telescoping in women who did not bleed by the time of the biopsy, (B) aborted secretions. Scale bar = 70 µm.

 
Irrespective of the dose, the mean total vascular space areas (Figure 2eGo) were smaller than that in the late luteal or menstrual phases (P < 0.0001). Similarly the average vascular space diameter (Figure 2fGo) was smaller than in the natural cycle (P < 0.001). However, there was no difference in the mean number of the vascular spaces in the HRT cycle compared to the natural cycle (Figure 2gGo). Stromal cell count did not differ between the four dose groups or in comparison to the natural cycle (Figure 2hGo).

There was no correlation between the histological parameters studied and the bleeding patterns in women who bled by the time of the biopsy or those who did not. Telescoping was more prevalent, however, in women who had not bled by the time of the biopsy compared to those who had already bled on that day (P < 0.002).

Linear discriminant analysis
Table IGo shows the mean scores for the 13 histological variables described above in the four treated groups and the five phases of the normal cycle. The within-group standard deviation and correlations between the endometrial parameters are shown in Table IIGo. The average number of vascular spaces, the mean total vascular area, and average vascular space diameter were highly correlated. High correlation was also found between the mean total glandular area and average glandular diameter. To highlight the differences between the groups a linear discriminant analysis was performed. The measurements were standardized by subtracting the overall mean from the individual observation value and dividing by the pooled within group standard deviations so that the effects of the scale of measurement were removed and the relative contributions of the different measurements were more clearly seen.


View this table:
[in this window]
[in a new window]
 
Table I. Mean (SD) values for individual histological parameters studied in the endometrium treated with sequential trimegestone and the natural cycle
 

View this table:
[in this window]
[in a new window]
 
Table II. The within-group standard deviation (SD) and correlations between individual histological parameters examined for the whole series of specimens
 
The linear discriminant functions (Table IIIGo) produced weighted combinations of the 13 measurements which best separated the nine groups (four doses of trimegestone, and five phases of the natural cycle). These functions were ordered so that the first linear discriminant function showed most discrimination, followed by the second, third, etc. The signs and sizes of the standardized weights show the importance of each variable to the final score. Thus mean total glandular area contributed highly to the score of linear discriminant function 1, while second linear discriminant function was associated with higher percentage of glands with vacuolation or telescoping, but lower mean vascular space area and lower percentage of glands with columnar epithelial lining.


View this table:
[in this window]
[in a new window]
 
Table III. Standardized weights assigned to individual histological parameters along the first four linear discriminant functions
 
Table IVGo shows the mean scores for women in the nine groups for each of the first four linear discriminant functions. The average scores for each group and scores for the individual women on the first two linear discriminant functions were plotted (Figure 4AGo).


View this table:
[in this window]
[in a new window]
 
Table IV. Mean scores on the first four linear discriminant functions, in the four dose groups of trimegestone compared to the natural cycle
 


View larger version (0K):
[in this window]
[in a new window]
 
Figure 4. The relationship between linear discriminant analysis scores of the histological parameters in the trimegestone-treated endometrium and the five phases of the natural cycle. Black crosses represent the mean scores of individual women in the treated group, and the red crosses represent group mean. Black circles represent mean scores of individual women in the control group and the green circles represent the group means. (A) This shows the distinction between the treated group and the luteal phases of the natural cycle along the first linear discriminant function, and the second linear discriminant function distinguishes between the trimegestone group and the proliferative, menstrual and mid luteal phases of the natural cycle. (B) Obtained from rotating (A) through 30° anti-clockwise. The first rotated linear discriminant function distinguishes between the treated group and the control, whereas the second rotated linear discriminant function distinguishes more clearly between all phases of the natural cycle. (C) Repeat plot of (A) using weighted scores >0.4 only. The discrimination between the groups is retained, as in (A); however, the proliferative phase is now less distinguished from the treated group. P = proliferative; EL = early luteal; ML = mid luteal; LL = late luteal; M = menstrual.

 
On the basis of these measurements, the trimegestone-treated endometria appeared to be very similar, regardless of the dose. The first linear discriminant function separated the luteal phases from the trimegestone and the second linear discriminant function separated the proliferative, menstrual, and mid luteal phases from the trimegestone-treated groups. Therefore, the averages for the trimegestone-treated groups could be distinguished from the averages of each of the phases of the natural cycle, although the variation between the women within these groups was substantial and there was considerable overlap.

If this first plot (Figure 4AGo) was rotated anti-clockwise through 30° to the horizontal axis, the first rotated linear discriminant functions then distinguished between the treated and control groups and the second rotated linear discriminant functions distinguished between the phases of the natural cycle (Figure 4BGo). The standardized weights for defining the rotated linear discriminant functions are shown in Table VGo.


View this table:
[in this window]
[in a new window]
 
Table V. Standardized weights for the rotated scales shown in Figure 4BGo
 
The rotated linear discriminant functions depended most critically on just a few variables. If only the variables with standardized coefficients >0.4 were used, then discriminant functions were obtained that captured most of the discriminatory information in this data set. The resulting plot is shown as Figure 4CGo. Most of the discriminatory information is retained although the proliferative phase is now less well distinguished from the treated groups.

Using the original raw data to reconstruct Figure 4CGo, the following equations could be used to assign values for the first and second linear discriminant functions, where they could be compared to the different phases of the natural cycle or to the trimegestone-based HRT regimen:


    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
The diagnosis of secretory endometrium in women treated with sequential combined HRT has been considered as the yardstick of adequacy of the progestogen administered. The mean day of onset of bleeding was considered to reflect the histological condition of the endometrium as a predictor of the adequacy of the progestogen effect in the cyclical sequential HRT regimen (Padwick et al., 1986Go). Women who bled before day 11 of the progestogen phase had a proliferative endometrium, hence this group recommended an increase in the dose of progestogen in these women. Nevertheless, the authors did not make it clear as to the proportion of these `early bleeders' who had secretory endometrium. In our study, the trimegestone dose-dependent mean day of onset of the progetogen-associated bleeding did not reflect the histological diagnosis as ~96% of the endometrial samples examined at the end of the study were secretory. This agrees with another group (Sturdee et al., 1994Go) who reported on the histological findings of endometrial samples obtained from 413 postmenopausal women using different HRT preparations, and found no correlation between the timing of the withdrawal bleeding and endometrial histology.

In the natural cycle, the mean total glandular area and average glandular diameter were higher in the luteal phase, the height of the glandular epithelium was lower in the mid luteal phase, while there was no difference in the number of glands, which agrees with the published literature (Johannisson et al., 1987Go, 1990).

The smaller mean total glandular area in this trimegestone-based HRT, compared to the natural cycle, agrees with the published literature on sequential HRT regimens; however, the average glandular diameter was smaller, while there was no difference in the average glandular number, which is at variance to the finding reported using a cyclical sequential oestradiol and norethisterone-based HRT regimen (Habiba et al., 1998Go). This may be due to the different progestin administered in this study which has no affinity to oestrogen receptors and minimal androgen receptor affinity, while norethisterone has potent androgen receptor affinity in addition to progesterone receptor binding; therefore, it may exert a different effect on glandular tortuosity with different number of sectioned glands appearing per unit area. Moreover, norethisterone, through its androgen receptor binding, may differentially affect the endometrium through the increase of epidermal growth factor (EGF) receptor expression, thus enhancing the action of EGF, which is involved in endometrial glandular growth and differentiation (Watson et al., 1998Go). Lastly, the possibility exists of the potential effect of conversion of norethisterone to ethinyl oestradiol (Fotherby, 1994Go), thereby augmenting the oestrogenic stimulation of the endometrium.

The highest dose of trimegestone (0.5 mg/day) was associated with lower glandular secretion. This may be similar to the aborted secretion (Dallenbach-Hellweg and Poulsen, 1996Go), which may indirectly explain the more favourable pattern of bleeding in this group in comparison to women on the lower doses (Al-Azzawi et al., 1999Go).

The assessment of the endometrial glands as a 3-dimensional model may theoretically provide more information on endometrial development (Mayhew and Gundersen, 1996Go). Such a method extrapolates the glandular dimensions, for example, from sections obtained at different levels of the tissue block, and as such adds further hypothetical assumptions which may reduce the power of the linear discriminant analysis.

Ota et al. studied the endometrial vascularity in the natural cycle, and found significant increase in the total vascular, number of capillaries and the mean diameter in the luteal phase compared to the proliferative phase, which is at variance with our findings (Ota et al., 1998Go). Casanas-Roux et al., on the other hand, studied the vascularization of the endometrium under the effect of cyclical sequential HRT using natural progesterone-containing vaginal gel (Casanas-Roux et al., 1996Go). They reported an increase in the capillary:stroma relative surface area in the HRT cycle. These findings were not confirmed in our study, as there was smaller mean total vascular space area and diameter in the HRT cycle compared to the late luteal and menstrual phases. This may be a special effect of trimegestone-based HRT, which, while having no androgen receptor binding activity, is not identical to the natural progesterone either.

There was no significant difference in the stromal cellularity between the five phases of the natural cycle, or between the four dose groups, which is at variance with other reports (Johannisson et al., 1987Go; Dockery et al., 1990Go). The latter group found an increase in the number of stromal nuclei per unit volume of stromal cells in LH + 2 to 6, but decreased in day LH + 6 to 8. However, Noyes et al. encountered a similar apparent paradox created by intercellular oedema, where stromal cells appear less in number due to the small, dense nuclei (Noyes et al., 1950Go). Stromal cellularity included the stromal cells and the infiltrating leukocytes, and we have documented that stromal leukocytes increase in the late luteal phase by ~149% (Figure 5Go, Wahab et al., 1999Go). Nevertheless, even if we allow for this increase, it will only have a minimal effect on the total stromal cell count. Indeed, in a separate analysis, the addition of the leukocyte common antigen (CD45+) cell count for the whole series of specimen failed to add a significant effect on linear discriminant functions (data not shown). To our knowledge there are no similar data on stromal cell counts in the published literature to which our finding can be compared.



View larger version (41K):
[in this window]
[in a new window]
 
Figure 5. The endometrial stromal cell count (grey bars) and the number of CD45+ cells (black bars) in the natural cycle, modified from Wahab et al. (1999).

 
In this study the only difference in the endometrium of women who bled on the day of the biopsy, compared to those who did not, was more glandular invagination (telescoping) in the latter group; which may be due to the fact that these glands have been shed during the bleeding. Many pathologists, nevertheless, argue that glandular telescoping is merely the result of mechanical disruption of the gland during fixation and cutting (Mazur and Kurman, 1995Go).

Apart from lower glandular secretion in the higher dose groups, there was no difference in the other histological parameters between the four dose groups. A retardation in the endometrial morphometry on the lower doses of progesterone has been reported (Li et al., 1992Go), although higher doses of progesterone did not accelerate or augment the secretory changes. They concluded that the advancement of the endometrial development was mediated by mechanisms other than the dose of progesterone.

Thus far, direct comparisons of individual histological parameters have failed to illustrate a determinant factor which characterizes the trimegestone-treated endometrium. Therefore, in the presence of a mixed picture of endometrial development under HRT regimens and the wide inter-specimen variability, we evaluated the relative weighting of these histological parameters by adopting linear discriminant analysis. To our knowledge this is the first study using this statistical method of analysis to evaluate differential changes in endometrial histomorphometry in the HRT cycle compared to the natural cycle.

In a large multivariate data set, it is frequently found that subsets of the variables reflect common underlying changes measured in slightly different ways, as for instance with glandular area and glandular diameter. Such underlying factors can sometimes be extracted by looking at weighted combinations of the measurements. In a linear discriminant analysis, weighted combinations of the measurements are sought that best separate out the different groups of subjects. By identifying those measurements that have the largest standardized weights, it is possible to tell which variables are important in distinguishing between the groups.

Linear discriminant analysis of the 13 histological variables examined helps to distinguish between the endometrium of the treated group and the five phases of the natural cycle. The endometrium of the natural cycle is distinguished from the trimegestone-treated endometrium in the first rotated linear discriminant function by larger mean total glandular and vascular areas, and lower percentage of glands with telescoping. The second rotated linear discriminant function distinguishes between the proliferative and the early luteal phases. The proliferative phase is characterized by larger mean total vascular space area but smaller average vascular diameter. Conversely the early luteal phase is associated with large mean total glandular area, and small average glandular diameter, in addition to higher percentage of glands with vacillations. As the mean number of the glands did not differ in the early luteal from the other phases of the natural cycle, the larger mean total glandular area with the smaller average glandular diameter is probably due to the increase in glandular tortuosity, which changes glandular shape.

By using the equations (i) and (ii), the construction of a position assigned to weighted observations in an endometrial specimen may serve to compare the effect of an HRT treatment on tissue morphometry to those of the natural cycle.

We realize that the trimegestone-treated endometrial samples obtained in our study represent a snapshot of histological changes at a fixed time point of the progestogen administered, day 10 (day 24 of this cyclical HRT cycle), but we needed to compare the extent of the progestogen effect in influencing endometrial differentiation. As such weighted values of histological parameters demonstrated in this study are dose independent, they may, therefore, represent either a specific effect of trimegestone, number of days administered, or both. In addition, one may speculate that such difference may be accounted for by a different oestrogen bioavailability to the endometrium in the individual woman; however, we feel it is less probable as all women received a fixed dose of oestrogen and such difference in bioavailability would be randomly distributed between the four dose groups. Moreover, the endometrial biopsy was taken on one fixed day of the treatment cycle.

This paper leads the way to potential starting points for investigation of a specific effect of a progestogen on endometrial compartments, and in the design of future more specific therapies.


    Acknowledgments
 
This study was supported by Hoechst Marion Roussel, Romainville, France and Wyeth-Ayerst, International, USA, in a joint development programme.


    Notes
 
4 To whom correspondence should be addressed Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Al-Azzawi, F., Wahab, M., Thompson, J. et al. (1999) Acceptability and patterns of uterine bleeding in sequential trimegestone-based HRT: a dose-ranging study. Hum. Reprod., 14, 636–641.[Abstract/Free Full Text]

Bouchoux, F., Cerede, E., Philibert, D. (1995) Measurement of the relative binding affinity of RU27987 to recombinant human steroid receptors: progestogen, glucocorticoid, androgen and oestrogen determination of the binding parameters of RU27987 and progesterone for the recombinant human progesterone receptor. 1995; 93/5353/PH data on file at Roussel-Uclaf.

Casanas-Roux, F., Nisolle, M., Marbiax, E. et al. (1996) Morphometric, immunohistological and three dimentional evaluation of the endometrium of menopausal women treated by oestrogen and crinone, a new slow-release vaginal progesterone. Hum. Reprod., 11, 357–363.[Abstract]

Dallenbach-Hellweg, G. and Poulsen, H. (eds) (1996) Histopathology of the Endometrium. Springer, New York, pp. 105 and 109.

Dockery, P., Warren, M.A., Li, T.C. et al. (1990) A morphometric study of the human endometrial stroma during the peri-implantation period. Hum. Reprod., 5, 494–498.[Abstract]

Fotherby, K. (1994) Pharmaco-kinetics and metabolism of progestins in humans. In Goldzieher, J.W. and Fotherby, K. (eds), Pharmacolcogy of the Contraceptive Steroids. Raven Press, New York. pp. 99–127.

Habiba, M.A., Bell, S.C. and Al-Azzawi, F. (1998) Endometrial responses to hormone replacement therapy: histological features compared with those of late luteal phase endometrium. Hum. Reprod., 13, 1674–1682.[Abstract]

Hamilton, P.W. (1995) Designing a morphometric study. In Hamilton, P.W. and Allen, D.C. (eds), Quantitative Clinical Pathology. Blackwell, Oxford, pp. 311–315.

Johannisson, E. (1990) Endometrial morphology during the normal cycle and under the influence of contraceptive steroids. In D'Arcangues, C., Fraser, I.S., Newton, J.R. and Odlind, V. (eds), Contraception and Mechanism of Endometrial Bleeding. Cambridge University Press, Cambridge, pp. 53–80.

Johannisson, E., Landgren, B.M. and Diczfalusy, E. (1982) Endometrial morphology and peripheral steroid concenrations in women with and without intermenstrual bleeding during contraception with the 300 µg norethisterone minipill. Contraception, 25, 13–30.[ISI][Medline]

Johannisson, E., Landgren, B. M., Rohr, H. P. and Diczfalusy, E. (1987) Endometrial morphology and peripheral hormone levels in women with regular menstrual cycles. Fertil. Steril., 48, 401–408.[ISI][Medline]

Johannisson, E., Brosens, I., Cornillie, F., Elder., M. et al. (1991) Morphometric study of the human endometrium following continuous exposure to levonorgestrel released from vaginal rings during 90 days. Contraception, 43, 361–371.[ISI][Medline]

Li, T.C., Ramsewak, S.S., Lenton, E.A. et al. (1992) Endometrial responses in artificial cycles: a prospective, randomised study comparing three different progesterone dosages. Br. J. Obstet. Gynaecol., 99, 319–324.[ISI][Medline]

Ludwig, H. (1982) The morphologic response of the human endometrium to long term treatment with progestational agents. Am. J. Obstet. Gynecol., 142, 796–808.[ISI][Medline]

Mardia, K.V., Kent, J.T. and Bibby, J.M. (1979) Multivariant Analysis. Academic Press, London.

Mayhew, T.M. and Gundersen, H.J.C. (1996) If you assume, you can make an ass out of u and me: a decade of the disector for stereological counting of particles in 3D space. J. Anat., 188, 1–15.[ISI][Medline]

Mazur, M. and Kurman, R.J. (1995) Diagnosis of Endometrial Biopsies and Curettings: A Practical Approach. Springer, New York, pp. 23.

Noyes, R.W., Hertig, A.T. and Rock, J. (1950) Dating the endometrial biopsy. Fertil. Steril., 1, 3–25.[ISI][Medline]

Ota, H., Igarashi, S. and Tanaka, T. (1998) Morphometric evaluation of stromal vascularization in the endometrium in adenomyosis. Hum. Reprod., 13, 715–719.[Abstract]

Padwick, M.L., Pryse-Davies, J. and Whitehead, M. I. (1986) A simple method for determining the optimal dosage of progestin in postmenopausal women receiving oestrogens. N. Engl. J. Med., 315, 930–934.[Abstract]

Sturdee, D.W., Barlow, D.H., Ulrich, L.G. et al. (1994) Is timing of withdrawal bleeding a guide to endometrial safety during sequential oestrogen-progesterone replacement therapy? UK Continuous Combined HRT study investigators. Lancet, 344 (8928), 979–982.[ISI][Medline]

Wahab, M., Thompson, J. and Al-Azzawi, F. (1999) The distribution of endometrial leucocytes and their proliferating markers in trimegestone treated postmenopausal women compared to the endometrium of the natural cycle: a dose ranging study. Hum. Reprod., 14, 1201–1206.[Abstract/Free Full Text]

Watson, H., Franks, S. and Bonney, R.C. (1998) Regulation of epidermal growth factor receptor by androgen in human endometrial cells in culture. Hum. Reprod., 13, 2585–2591.[Abstract]

Submitted on June 14, 1999; accepted on July 19, 1999.