Hum. Reprod. Advance Access originally published online on January 23, 2008
Human Reproduction 2008 23(4):912-918; doi:10.1093/humrep/dem418
Why does menopausal hormone therapy lead to irregular uterine bleeding? Changes to endometrial blood vessels
1 School of Women's and Infants' Health, King Edward Memorial Hospital, University of Western Australia, Subiaco, Perth, Western Australia 6008, Australia 2 Sydney Centre for Reproductive Health Research, F.P.A. Health and Department of Obstetrics and Gynaecology, University of Sydney, Australia 3 Prince Henry’s Institute of Medical Research, Melbourne, Victoria, Australia 4 The Liggins Institute, The University of Auckland, Private Bag 92019, Auckland, New Zealand
5Correspondence address. Tel: +61-8-9340-1330; Fax: +61-8-9381-3031; E-mail: mhickey{at}meddent.uwa.edu.au
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
BACKGROUND: Abnormal bleeding is common in hormone therapy (HT) users. We aimed to determine how HT alters endometrial blood vessels and stromal factors known to regulate vascular growth and integrity.
METHODS: Prospective observational study of 165 post-menopausal women in Western Australia. The following were measured in endometrial biopsies: vascular density (vessels/mm2), total vessel area (total area enclosed by peripheral vascular immunostaining for perivascular pericytes in mm2), total luminal area (mm2) and vessel wall area (total vessel area minus luminal area), stromal expression of matrix metalloproteinases (MMP) -1, -3, -9 and -14, their tissue inhibitors (TIMPs) -1-4 and vascular endothelial growth factor (VEGF) by immunohistochemistry.
RESULTS: Total vessel area was greater during bleeding compared with HT users with no bleeding (P = 0.028) or with a prior irregular bleeding (P = 0.039). Total vessel area was greater in non-HT users compared with HT users with no bleeding (P = 0.021). In HT users, vessel luminal area was greater during bleeding compared with HT users with no bleeding (P = 0.030) and vessel wall area was also increased (P = 0.025). During bleeding there was an increase in stromal TIMP-2 staining (P = 0.044). No significant changes in endometrial MMP or VEGF were seen.
CONCLUSIONS: Abnormal bleeding in HT users is associated with changes in endometrial vessel size and in stromal expression of factors known to regulate vascular growth and integrity. These changes may contribute to abnormal bleeding.
Key words: menopause/irregular bleeding/hormone therapy/endometrium/blood vessels
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
Relatively few women go through menopause without experiencing vasomotor symptoms. For those who require treatment, estrogen containing hormone therapy (HT) remains the most effective therapy (MacLennan et al., 2001
). Irregular vaginal bleeding affects nearly half of standard dose HT users (Christodoulakos et al., 2006). Abnormal bleeding led to 40% unblinding in the Women’s Health Initiative (Rossouw et al., 2002) and is a common reason for discontinuation of HT. Concerns that bleeding may reflect underlying malignancy commonly lead to invasive and expensive investigations. Half of all HT users undergo at least one endometrial biopsy, ultrasound or hysteroscopy, but usually no pathology is found (Elliott et al., 2003). Management is often unsatisfactory, since there are no established methods of investigating, predicting, regulating or reducing bleeding.
The mechanisms of this bleeding are poorly understood and do not correlate well with endometrial histology or the type or dose of HT used (Thomas et al., 2000). We have previously shown increased uterine natural killer cells and an alteration in endometrial expression of matrix metalloproteinases (MMP) relative to the tissue inhibitors of metalloproteinases (TIMPs) (Hickey et al., 2005, 2006) associated with bleeding. These changes may also alter vascular density and morphology, which have not yet been systematically studied in HT users.
Endometrial bleeding must by definition involve simultaneous breakdown of endometrial vessels and their overlying epithelium. The aim of this study was to determine (i) the effect of combined HT exposure on endometrial vascular density and morphology, (ii) the relationship between these changes and bleeding patterns and (iii) the role of perivascular pericytes and key angiogenic and inflammatory agents known to alter endometrial vascular integrity.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
The study was approved by the institutional ethics committee of King Edward Memorial Hospital (KEMH, Western Australia) and all subjects gave written informed consent.
One hundred and sixty-five post-menopausal women were recruited at KEMH between 2003 and 2005, including 30 post-menopausal non-HT users with at least 12 months amenorrhoea and 135 users of combined oral or transdermal HT. Bleeding patterns were prospectively classified as ‘no bleeding’, ‘regular bleeding’ (cyclic HT only) or ‘irregular bleeding’. Irregular bleeding was investigated according to standard hospital protocols. Endometrial biopsies were collected using a Pipelle curette (Cornier, France) plus hysteroscopy when endometrial thickness was >6 mm and were fixed in 10% formalin for 6 h, washed in phosphate-buffered saline and paraffin embedded for future analyses.
Immunohistochemistry
A section from each biopsy was classified according to Noyes criteria (Noyes et al., 1975) by an experienced gynaecological pathologist. The number of subjects and biopsies are detailed in Table I. Immunohistochemical detection was performed on 5 µm sections using primary antibodies (Table II) with avidin-biotin-peroxidase reagents (Vectastain ABC Kit, Vector Laboratories, CA, USA) as described previously (Hickey et al., 2005). Primary antibody was substituted on adjacent sections by an iso-type non-specific immunoglobulin at the same concentration to provide negative controls.
|
|
Analysis of endometrial morphology and vasculature and protein localization
Analyses were performed using computerized image analysis (Image Pro Plus software; Media Cybernetics Inc, MD, USA) as described previously (Hickey et al., 2005
). Areas were selected where each field had at least 90% stroma (<10% gland or epithelium). All analyses (except vascular morphology) were performed by a single experienced observer. Individual immuno-labelled cells were counted within each measured field using a grid overlay (50 x 50 µm2) on each image. Stromal cell density was recorded using the same images by counting all haematoxylin stained nuclei in five separate fields, (50 x 50 µm), including those nuclei that crossed two sides only of each selected area, as specified for stereological counting.
Vascular density
CD34-positive endothelial cells organized as a collection of single cells grouped around a lumen were recorded as blood vessels. Vascular density was determined by counting the number of blood vessels within a 50 x 50 µm overlay grid and expressed as the median number of vessels per field, and total vessels per biopsy.
Total blood vessel area and luminal area
Perivascular pericytes are the most peripheral cells surrounding endometrial blood vessels. Total vessel area was estimated by determining the total area outlined by peripheral vascular immunostaining of smooth muscle actin (SMA)-positive pericytes. The lumen area was determined by measuring the area outlined by peripheral immunostaining of endothelial cells. These measurements were made using computerized image analysis (Image Pro Plus) by manually tracing the outline of SMA-positive cells and the outline of CD34-positive cells for each blood vessel within the overlaying grid. The ratio of total luminal area to total vessel area was calculated (Fig. 1).
|
Vessel wall area
In order to estimate the vessel wall area, the total luminal area (CD34 positive) was subtracted from the total vessel area (SMA positive) and expressed as vessel wall area in mm2 (Fig. 1).
MMPs and VEGF
Overall levels of MMPs, TIMPs -1-4 and vascular endothelial growth factor (VEGF) were determined using a semi-quantitative scoring system as described previously (Hickey et al., 2006). Immuno-positive staining for MMP-1, -3, -9, -14, TIMP-1-4 and VEGF were categorized on each section as: 0 = no labelled cells; 1 = occasional immuno-positive cells; 2 = moderate number of immuno-positive cells and 3 = large number of immuno-positive cells. The percentage of samples in each group having a score of 2 or 3 was defined as ‘% high’ staining for subsequent comparison between groups.
Statistical analysis
Descriptive statistics utilized medians and interquartile ranges (IQR) (Q1–Q3) and ranges (min–max) to describe continuous data. Stromal areas occupied by blood vessels were analysed with respect to the differences between HT groups. Total vessel area and luminal area were analysed after log-transformation of data due to lack of normal distribution of raw data. Mixed modelling with HT/bleeding status was formulated as a fixed effect and individual women were modelled as random effects. Mixed modelling was implemented to account for correlated responses with repeated biopsies per woman, either within or between different HT/bleeding status. All observations per woman were suitably weighted so that under each HT/bleeding condition the data for each woman carried equal weighting. The effects were summarized using means and their 95% confidence intervals obtained on a logarithmic scale and reported after back-transformation. All hypothesis tests were two-sided. P-values <0.05 were considered to be statistically significant and P-values for pairwise comparisons between groups were obtained using Tukey–Kramer correction for multiple testing. Data were analysed using SAS version 8.2 for Windows (SAC Inc, Cary IL, USA) statistical software.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
Two hundred and five biopsies were obtained from 165 women of which only 78 biopsies (38%) from 54 (33%) women were adequate for analysis. One biopsy was obtained from 36 subjects and >1 biopsy from 18 subjects. Biopsies were separated into four groups depending on current HT use and bleeding pattern at the time that the biopsy was taken (Table I) Adequate first biopsies were obtained from 8 women in group 1 (no HT and no bleeding), 6 women in group 2 (HT use and no bleeding), 22 women in group 3 (HT and irregular bleeding) and 18 women in group 4 (HT use and bleeding at the time of biopsy).
During the course of the study, a further 24 usable biopsies were obtained from these subjects, making 78 usable biopsies in total. These subsequent biopsies were classified according to the HT and bleeding status of the subject at the time that the biopsy was taken (i.e. zero in group1, 2 in group 2, 16 in group 3 and 6 in group 4). This reflects the clinical situation where women on HT experience changes in their bleeding patterns over time. The majority of repeat biopsies from the same subject were obtained under different conditions of HT use/bleeding patterns. For example, a subject may have had a biopsy taken as a HT user with no history of irregular bleeding in the previous 3 months (group 2). She might subsequently develop irregular bleeding and have a further biopsy which would be classified as group 3 (HT use and irregular bleeding in the previous 3 months, Table I). Bleeding which immediately followed an endometrial biopsy was not considered. In data analysis, subjects and biopsies have been weighted to account for the contribution of each individual to the study.
Of the 78 usable biopsies obtained, 8 were from post-menopausal women not taking HT (10%), 61 were using continuous combined HT (78%) and 9 were using cyclic HT (12%). All biopsies from cyclic HT users were obtained during estrogen and progestogen treatment days. The median age of HT users (54 years, IQR 50–57, range 42–64 years) and non-HT users (51 years, IQR 46–57, range 42–65 years) did not differ (P = 0.218). As anticipated, subjects were using a range of HT products (Table III).
|
Investigation of irregular bleeding
All Pap smears were negative. Ten subjects were excluded for polyps or submucous fibroids. No endometrial hyperplasia or cancers were identified.
Endometrial histology
Most biopsies were classified as either atrophic or weakly proliferative. The remainder were decidualized (8) or secretory (8). There was no significant relationship between endometrial histological appearance and bleeding patterns or the type of HT used.
Endometrial expression of MMPs, TIMPs and VEGF
There were no significant differences in endometrial expression of MMP-1, -3, -9 and -14 between the groups studied. There was a trend for TIMPs 1–4 to be increased in biopsies obtained during a bleeding episode (group 4), but this reached significance (P = 0.044) only for TIMP-2 (Table IV and Fig. 2). TIMP immunostaining was localized to the stroma as previously described (Hickey et al., 2006). No differences in endometrial VEFG were seen in association with HT use or bleeding patterns.
|
2 levels of each protein).
|
Endometrial vascular density
In total, 901 blood vessels were examined: 103 in group 1, 81 in group 2,440 in group 3 and 277 vessels in group 4. Median numbers of fields, vessels per field and total vessels per biopsy are summarized in Table V. Median vascular density did not differ between non-HT users and HT users (P = 0.947), or according to bleeding patterns in HT users (P = 0.968, Table V, Fig. 3).
|
|
Total vessel area
There were significant differences in total vessel area between HT users and non-users and according to bleeding patterns in HT users (Table V, P = 0.009, Fig. 4A). Endometrial vessels in HT users during bleeding (group 4) were significantly larger than in those with bleeding during the last 3 months (group 3, P = 0.039), and those with no bleeding (group 2, P = 0.028). These larger vessels appeared to be close to the endometrial surface, although this observation was not quantified. In the small number of usable biopsies from non-HT users (group 1) vessels were significantly larger than those seen in HT users without bleeding (group 2, P = 0.021, Fig. 4A).
|
Vessel luminal area
Vessel luminal area was variable between groups (P = 0.014, Fig. 4B). In HT users, luminal area was significantly greater during bleeding (group 4) compared with no bleeding (group 2, P = 0.030). In non-HT users, luminal area was similar to those biopsied during bleeding (group 4 P = 0.995) and greater than those of HT users with no bleeding (group 2 P = 0.033). There was no correlation between vascular luminal area and vascular density, MMP, TIMP or VEGF expression (data not shown). No significant differences in endometrial vascular density or vascular morphology and stromal expression of VEGF were seen between groups (data not shown).
Vessel wall area
Vessel wall area differed between groups (Table V, P = 0.005, Fig. 4C). In HT users, vessel wall area was significantly increased in those who were currently bleeding (group 4) (Fig. 4C, P = 0.025). It was also significantly greater in non-HT users (group 1) than in HT users with no bleeding (groups 2 and 3) (P = 0.024 and 0.018, respectively, Figs 4C and 5).
|
No relationship was seen between vessel wall area and vascular luminal area or density, MMP, TIMP or VEGF expression (data not shown).
Ratio of luminal area to total vessel area
Vessels with large lumens and reduced vessel wall area were indicated by high lumen to total vessel area ratios. Vessels with small lumens and increased vessel wall area were indicated by lower lumen to total vessel area ratios. There was an overall difference between groups (P = 0.028). HT users with a history of irregular bleeding had a marginally higher ratio of luminal area to total vessel area compared with non-HT users (P = 0.045, Fig. 4D).
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
The main finding of this study was that total vessel area and luminal area were elevated during bleeding in biopsies from HT users. This may reflect differences in endometrial vascular function. Premenopausal progestogen users with breakthrough bleeding (BTB) have enlarged, fragile vessels on the endometrial surface (Hickey et al., 2000
). We now report increased vessel size during bleeding, although it is not possible to say whether this caused the bleeding or resulted from it. We were surprised to find that vessel size in non-HT users was similar to HT users. We expected that vessel area would be increased during bleeding in HT users, but did not expect to find that vessel area during bleeding was similar to that seen in non-HT users. We expected to find smaller vessels in non-HT users, but observed that the smallest vessels were seen in HT users with no bleeding. These results suggest that the size of the vessels is unlikely to be responsible for the bleeding and is only part of the mechanisms underlying HT-related irregular bleeding. This observation also suggests a difference in mechanism between HT-related irregular bleeding and that seen with long-acting progestogen-only contraceptives (Hickey and Fraser, 2000). However, since very few usable samples were obtained from non-HT users these findings may not be representative. It is very difficult to obtain endometrial samples of adequate size for analysis from post-menopausal women who are not bleeding. Large samples can be obtained from hysterectomy specimens, but these subjects are likely to differ in age, time since menopause and clinical presentation compared with the ‘average’ HT user (Hickey et al., 2001). Biopsies from premenopausal women are larger, but the endogenous and exogenous hormonal milieu is completely different. A further limitation of this study is that the proportion of usable biopsies is different between the groups studied (range from 19% in group 2 to 52% in group 3). This is a potential source of bias since it cannot be assumed that subjects from which sufficient tissue was not obtained are the same as those from whom sufficient tissue was obtained. A challenge of any endometrial studies in post-menopausal women is obtaining adequate tissue for analysis.
Consistent with our previous findings, MMP production was not increased during bleeding (Hickey et al., 2001, 2006). Again, this suggests a different mechanism from both BTB and normal menstrual bleeding (Galant et al., 2004). We confirm our previous observation that TIMP-2 is increased during HT bleeding (Hickey et al., 2006). However, in this population the findings for the other TIMPs did not reach statistical significance. This suggests that the endometrial balance between MMP and TIMP may play a role in regulating bleeding on HT, but the precise nature of this role is not yet clear. Proteolytic activity of MMP in the endometrium is partly regulated by TIMPs which bind active MMPs and prevent their action. TIMPs also act independently of MMP to regulate cell growth (Murphy et al., 1993). TIMPs are also increased at the time of normal menstruation (Zhang et al., 1997). It may be that the observed increase in TIMPs is a reaction to, rather than a cause of, the bleeding.
We were surprised to find that vessel wall area was increased during bleeding. Previous studies with progestogen users have shown enlarged, thin walled vessels on the endometrial surface in subjects with irregular bleeding (Runic et al., 2000), although the precise relationship between these vessels and bleeding episodes is less clearly defined. We have used the area immunostaining for perivascular pericytes as a marker of vessel wall thickness and perivascular supporting structures. Pericytes surround capillaries during maturation, contribute to basement lamina formation and act to strengthen and stabilize vessels. We have previously shown that perivascular pericytes are reduced in HT users with a history of irregular bleeding (Hickey et al., 2003). Our findings suggest that while vascular integrity might be reduced in HT users with irregular bleeding, pericytes may also contribute to vascular repair.
In summary, we have demonstrated for the first time a relationship between the size of endometrial vessels and bleeding in HT users, and changes in perivascular pericytes during bleeding episodes. These stromal changes differ from those seen in premenopausal women with BTB bleeding from an atrophic endometrium. It is unlikely that effective mechanisms for the treatment or prevention of abnormal bleeding on HT will be developed unless the underlying mechanisms are understood. Our findings provide novel and potentially clinically important new insights into why combined HT commonly induces abnormal bleeding.
|
Funding |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
M.H., I.S.F. and L.A.S. are supported by National Health and Research Council of Australia grants (#388901, #254645).
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
We acknowledge the support and technical help of the KEMH Pathology Department staff and particularly Barbara Brennan for expert evaluation of the histological specimens. Also, Julie Crewe for performing the immunohistochemistry and Lee-Ann Mahoney for subject recruitment.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
Christodoulakos GE, Botsis DS, Lambrinoudaki IV, Papagianni VD, Panoulis CP, Creatsa MG, Alexandrou AP, Augoulea AD, Dendrinos SG, Creatsas GC. A 5-year study on the effect of hormone therapy, tibolone and raloxifene on vaginal bleeding and endometrial thickness. Maturitas (2006) 53:413–423.[CrossRef][Web of Science][Medline]
Elliott J, Connor ME, Lashen H. The value of outpatient hysteroscopy in diagnosing endometrial pathology in postmenopausal women with and without hormone replacement therapy. Acta Obstet Gynecol Scand (2003) 82:1112–1119.[CrossRef][Web of Science][Medline]
Galant C, Berliere M, Dubois D, Verougstraete JC, Charles A, Lemoine P, Kokorine I, Eeckhout Y, Courtoy PJ, Marbaix E. Focal expression and final activity of matrix metalloproteinases may explain irregular dysfunctional endometrial bleeding. Am J Pathol (2004) 165:83–94.
Hickey M, Fraser IS. A functional model for progestogen-induced breakthrough bleeding. Hum Reprod (2000) 15:1–6.
Hickey M, Dwarte D, Fraser IS. Superficial endometrial vascular fragility in Norplant users and in women with ovulatory dysfunctional uterine bleeding. Hum Reprod (2000) 15:1509–1514.
Hickey M, Higham J, Sullivan M, Miles L, Fraser IS. Endometrial bleeding in hormone replacement therapy users: preliminary findings regarding the role of matrix metalloproteinase 9 (MMP-9) and tissue inhibitors of MMPs. Fertil Steril (2001) 75:288–296.[CrossRef][Web of Science][Medline]
Hickey M, Pillai G, Higham JM, Sullivan M, Horncastle D, Doherty D, Stamp G. Changes in endometrial blood vessels in the endometrium of women with hormone replacement therapy-related irregular bleeding. Hum Reprod (2003) 18:1100–1106.
Hickey M, Crewe J, Goodridge JP, Witt CS, Fraser IS, Doherty D, Christiansen FT, Salamonsen LA. Menopausal hormone therapy and irregular endometrial bleeding: a potential role for uterine natural killer cells? J Clin Endocrinol Metab (2005) 90:5528–5535.
Hickey M, Crewe J, Fraser IS, Doherty D, Salamonsen L. Mechanisms of irregular bleeding with hormone therapy: the role of matrix metalloproteinases and their tissue inhibitors. J Clin Endocrinol Metab (2006) 91:3189–3198.
MacLennan A, Lester S, Moore V. Oral estrogen replacement therapy versus placebo for hot flushes: a systematic review. Climacteric (2001) 4:58–74.[Medline]
Murphy AN, Unsworth EJ, Stetler-Stevenson WG. Tissue inhibitor of metalloproteinases-2 inhibits bFGF-induced human microvascular endothelial cell proliferation. J Cell Physiol (1993) 157:351–358.[CrossRef][Web of Science][Medline]
Noyes RW, Hertig AT, Rock J. Dating the endometrial biopsy. Am J Obstet Gynecol (1975) 122:262–263.[Medline]
Rossouw JE, Anderson GL, Prentice RL, LaCroix AZ, Kooperberg C, Stefanick ML, Jackson RD, Beresford SA, Howard BV, Johnson KC, et al. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results From the Women’s Health Initiative randomized controlled trial. JAMA (2002) 288:321–333.
Runic R, Schatz F, Wan L, Demopoulos R, Krikun G, Lockwood CJ. Effects of norplant on endometrial tissue factor expression and blood vessel structure. J Clin Endocrinol Metab (2000) 85:3853–3859.
Thomas AM, Hickey M, Fraser IS. Disturbances of endometrial bleeding with hormone replacement therapy. Hum Reprod (2000) 15(Suppl 3):7–17.
Zhang J, Salamonsen LA. Tissue inhibitor of metalloproteinases (TIMP)-1, -2 and -3 in human endometrium during the menstrual cycle. Mol Hum Reprod (1997) 3:735–741.
Submitted on January 1, 2007; resubmitted on December 5, 2007; accepted on December 13, 2007.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||