| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
School of Women’s and Infants’ Health (M.H., J.C., L.A.M., D.A.D.), University of Western Australia, Perth, Australia 6008; Sydney Centre for Reproductive Health Research (I.S.F.), Family Planning Association Health and Department of Obstetrics and Gynaecology, University of Sydney, Sydney, Australia 2006; and Prince Henry’s Institute of Medical Research (L.A.S.), Melbourne, Victoria, Australia 5152
Address all correspondence and requests for reprints to: Associate Professor Martha Hickey, School of Women’s and Infants’ Health, King Edward Memorial Hospital, Subiaco, Perth, Western Australia 6008. E-mail: mhickey{at}obsgyn.uwa.edu.au.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Objective: The aim of this study was to determine how HT exposure alters endometrial production of MMP-1, -3, -9, and -14 and their tissue inhibitors TIMP-1, -2, -3, and -4 and to determine the relationship between MMP and TIMP production and bleeding patterns in HT users. Endometrial leukocytes regulating MMP production and activation were also assessed.
Design: A prospective observational study was conducted between 2003 and 2005.
Setting and Patients: The study occurred at a tertiary referral menopause clinic at King Edward Memorial Hospital, Western Australia, and included 25 postmenopausal women not taking HT and 73 women taking combined HT.
Interventions: Endometrium was obtained during and outside bleeding episodes.
Main Outcome Measures: We assessed production of MMP-1, -3, -9, and -14 and their tissue inhibitors TIMP-1, -2, -3, and -4 and their relationship to bleeding patterns in HT users.
Results: All MMPs studied, with the exception of MMP-9, were expressed at low levels in postmenopausal endometrium. Increases in both MMP-3 and -9 localization were seen in association with irregular bleeding, but these did not reach statistical significance. Endometrial production of TIMP-1 was significantly increased in association with bleeding. Endometrial leukocytes were not related to bleeding, with the exception of uterine natural killer cells, which were significantly increased during bleeding, as previously published.
Conclusions: Irregular bleeding in HT users is associated with a distinct pattern of MMP and TIMP production that differs from that seen in normal menstrual bleeding and from that seen in contraceptive-related breakthrough bleeding. This suggests that the endometrial balance between MMP and TIMP contributes to vascular breakdown with HT but by a different mechanism than that seen in normal menstruation or in breakthrough bleeding.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Irregular bleeding is common to all regimens and preparations, suggesting that common mechanisms underlie these phenomena, independent of the precise quantities and timings of hormone provision (6). The management of irregular bleeding on HT is often unsatisfactory, because there are no established methods of regulating or reducing bleeding.
The mechanisms of HT-induced bleeding are poorly understood. We (and others) have previously demonstrated that bleeding patterns do not correlate with endometrial histology or the type or dose of HT used (7, 8). We have previously demonstrated low production of matrix metalloproteinase (MMP)-9 in the endometrium of HT users, but a change in the endometrial balance between MMP-9 and its tissue inhibitor of metalloproteinase (TIMP)-1 compared with postmenopausal women not using HT (9), that endometrial blood vessels density is reduced in HT users with reduced vascular support (7), and that endometrial uterine natural killer (uNK) cell populations are increased in association with bleeding (10).
Normal menstrual bleeding shows many similarities to inflammation with a tightly regulated influx of leukocytes preceding endometrial breakdown and bleeding (11). Leukocytes are a source of MMP that contributes to bleeding via degradation of components of the vascular basement lamina (12). Endometrial production and activation of MMP-1, -3, -9, and -14 varies cyclically (13, 14) and is strongly increased after withdrawal of exogenous or endogenous progestogen (15). Critical to the regulation of MMP activity is inhibition by TIMPs, which form 1:1 complexes with the active enzymes. Altered endometrial MMP-9 (16), MMP-3 (17), and TIMP-1 (18) production and leukocyte populations (19, 20) are seen in premenopausal women using progestogen-only contraceptives who complain of breakthrough bleeding (BTB). Apart from our pilot studies showing alterations in the ratio of MMP-9 to TIMP-1 in postmenopausal women after HT exposure, the potential role of MMP in HT-related bleeding has not previously been studied. Similarly, apart from our observations with uNK cells, the role of leukocytes has not previously been explored. This study further defines the role of other key MMPs, TIMPs, and their regulating leukocytes in HT users and adds significant and novel information about underlying mechanism regulating bleeding.
There is a striking variability in bleeding patterns between women exposed to the same hormonal regimen. It is not known whether this variation reflects endometrial, systemic, or other factors, but further understanding of this may hold the key to successful clinical intervention or prevention. We have investigated the sources of variation in bleeding patterns in HT users by comparing endometrium from users with irregular bleeding to those with amenorrhea.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Biopsies were fixed immediately in 10% formalin for 18 h and then embedded in paraffin blocks using a standard dehydration protocol. Sections were cut at 5 µm on a Leica RM2135 microtome on to SuperFrostPlus glass slides (Menzel-Glaser, Strasburg, Germany) and dried at 37 C overnight. All biopsies had a standard hematoxylin and eosin section submitted to an experienced gynecological pathologist for classification according to Noyes criteria (22).
Immunohistochemical labeling
Sections were rehydrated through three changes of Histoclear (National Diagnostics, Atlanta, GA) and descending concentrations of ethanol to distilled water followed by 0.01 M PBS (pH 7.4). Antigen retrieval was carried out by exposing the sections to 0.01 M trisodium citrate buffer (pH 6) at a temperature over 100 C for 10 min in a domestic microwave oven. Endogenous peroxidase was blocked by immersing the sections in 3% hydrogen peroxide in PBS for 10 min. After thorough rinsing in three changes of PBS, the slides were incubated in a blocking solution containing 1% BSA (Sigma Chemical Co., St. Louis, MO), 0.01% Tween 20 (Sigma), and 1% nonspecific horse serum. After PBS rinsing, the tissue sections were incubated overnight at 4 C with primary antibody as detailed in Table 1
.
|
Analysis of immunostaining
Cell counts per square centimeter of stroma and per 1000 stromal cells were obtained for leukocyte populations (CD56+ uNK cells, as previously published, elastase+ polymorphic neutrophils, CD3+ T cells, and CD68+ macrophages) by photographing multiple fields from each immunolabeled biopsy section, using a x40 objective lens plus x10 eye piece on an Olympus BX51 light microscope fitted with a SPOT RT Slider digital camera (Diagnostic Instruments, Sterling Heights, MI). Images were obtained from five separate fields at x400 magnification, maximizing stromal tissue and limiting gland and epithelial content to less than 10% of the field area. All images were analyzed for specific labeled cells using the image analysis program Image-Pro Plus (Media Cybernetics Inc., Silver Spring, MD). Individual labeled cells were counted within each measured field (200 x 250 µm, 0.05 mm2) using a grid overlay on each image. Stromal cell density was recorded using the same images by counting all hematoxylin-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. Mean number of immunolabeled cells, the stromal cell density per square millimeter, and final number of labeled leukocytes/1000 stromal cells were calculated.
All other labeling (MMPs, TIMPs, tryptase+ mast cells, and CD45+ B cells) was evaluated by a single observer (J.C.) who repeated the scoring of each anonymous immunolabeled biopsy section twice, on separate days, using the following system to categorize the immunolocalization: 0 = nil labeled cells, 1 = occasional positive labeled cells present, 2 = moderate number of positive cells present, 3 = high numbers of labeled cells present (10). The overall level of MMP or TIMP localization within each biopsy section was scored rather than individual cells counted because these proteins were usually located in foci and often protein was distributed extracellularly, making cell counting inaccurate.
Statistical analysis
Sample sizes were determined by an a priori power calculation with greater than 80% power to detect differences in MMP production of 50% between the groups for binary outcomes and greater than 90% power to detect differences of more than 1.5 SD for continuous outcomes.
Descriptive statistics for continuous data used medians and ranges because of lack of normality. Categorical data were summarized using frequency distributions. Production of MMP and TIMP data were categorized according to no/weak production (scores 0–1) and moderate/strong production (scores 2–3), as appropriate for semiquantitative ordinal data (23). Ages of HT users and nonusers were compared using Mann-Whitney U test. Differences between HT groups were analyzed using linear and logistic mixed models (23) where HT/bleeding groups were modeled as fixed effects to evaluate group differences and individual women were modeled as random effects to account for multiple biopsies sampled per woman. All observations per woman were suitably weighted so that under each HT/bleeding condition, the data for each woman had an equal contribution to the overall effect. Linear regression was used to compare continuous data, and logistic regression was used to compare categorical outcomes such as MMP and TIMP production. In comparisons of categorical (binary) outcomes, direct group comparisons and tests for trend were performed using a likelihood ratio test based on a 
2 distribution. When appropriate, pairwise group contrasts were performed using a Wald test based on a 2 distribution. Cell counts for leukocytes were analyzed as log-transformed to normalize the data because of lack of normality of distribution in the raw counts. SAS (version 8.02; SAS Institute, Cary IL) was used for data analysis. All hypothesis tests were two sided, and P values < 0.05 were considered statistically significant.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
An a priori decision was made to analyze subjects in four categories to reflect clinical differences in HT exposure and bleeding patterns. Group 1 included postmenopausal women (no bleeding for 12 months) and no HT use in the previous 3 months (25 women biopsied and eight adequate biopsies obtained). Group 2 was postmenopausal women taking HT for more than 3 months with no bleeding (26 women, nine adequate biopsies obtained). Group 3 included postmenopausal women taking HT for more than 3 months with irregular bleeding (57 women, 41 adequate biopsies obtained), and group 4 was the same as group 3, but biopsies were obtained during a bleeding episode (32 women, 25 adequate biopsies obtained) (Table 2).
|
As anticipated, subjects were using a range of HT products (see Table 3). All subjects in the longitudinal study were commenced on the same oral HT preparation (Kliovance; Novo Nordisk, New South Wales, Australia) containing 1 mg micronized estradiol and 0.5 mg norethindrone acetate.
|
Pap smears were performed if a normal Pap smear was not available for the previous 12 months. All results were negative. Fifteen subjects had an endometrial thickness of more than 6 mm on transvaginal ultrasound and underwent hysteroscopy. Seven of these women were then excluded because of endometrial pathology (polyps or pedunculated submucous fibroids). No endometrial hyperplasia or cancers were identified.
Endometrial histology
Biopsies were classified as being either atrophic or weakly proliferative in most cases, and the remainder were decidualized (eight) or secretory. The eight decidualized biopsies were found in one (11%) of nine HT users with no bleeding for more than 3 months (group 2), three (7%) of 41 biopsies from HT users with irregular bleeding (group 3), and four (16%) of 25 biopsies taken at the time of bleeding (group 4). Tissue breakdown as described by Galant et al. (24, 25) was specifically identified in 13 biopsies, all of which had been taken from women with irregular bleeding on HT (group 3), and six (46%) of 13 were from biopsies taken during a bleeding episode (group 4).
No significant relationship was seen between endometrial histological appearance and bleeding patterns or the type of estrogen or progestogen in the HT used. No consistent differences were observed between cyclic and continuous combined HT users in endometrial histological appearance.
Immunohistochemistry
The immunohistochemical results are summarized in Table 4.
|
No statistically significant increase in MMP-9 immunostaining was seen between the users and nonusers of HT or between those with bleeding and those with no bleeding. Immunolocalization of the other MMPs examined, MMP-1, -3, and -14, was at consistently low levels and was also unrelated to bleeding patterns.
MMP-1 was undetectable in about one third (34%) of biopsies and was seen at only very low levels in almost all others. MMP-1 was located in small isolated foci of 10–20 stromal cells and occasionally adjacent to glands (Fig. 1A). Occasionally, MMP-1 production was seen in adjacent glandular epithelial cells but was not associated with blood vessels and did not appear to relate to bleeding patterns.
|
), which appeared from their hematoxylin and eosin staining properties to be leukocytes and were weakly present in the cytoplasm of larger decidualized stromal cells. There was a nonsignificant trend toward increased MMP-3 production in HT users with a history of irregular bleeding (P = 0.06, 2 test for trend).
MMP-9 was produced in most biopsies by vascular, glandular, and surface epithelial cells and most markedly in the cytoplasm of stromal cells and leukocytes (Fig. 1C). The most prominent localization of MMP-9 was seen in decidualized stromal cells (Fig. 1D). In HT users with no bleeding, MMP-9 immunolocalization was limited to epithelial cells and to small blood vessels. Endometrial MMP-9 production was not significantly related to bleeding patterns in HT users.
Low-level immunolocalization of MMP-14 was seen in the glandular epithelium and occasionally in stromal cells (Fig. 1E). MMP-14 immunolocalization was not related to bleeding patterns.
TIMPs
There was a statistically significant relationship between endometrial production of TIMP-1 and irregular bleeding in HT users. The percentage of women with strong endometrial immunolocalization of TIMP-1 was significantly different between the four groups (P = 0.04) and was significantly increased in those with irregular bleeding compared with those with amenorrhea (Fig. 2, A and B). The most prominent TIMP-1 production was seen in biopsies obtained during a bleeding episode (P = 0.02). There was a nonsignificant trend toward elevated TIMP-1 production in biopsies taken from subjects with a history of irregular bleeding compared with those with no bleeding (P = 0.06). A similar trend was seen with TIMP-2 production, but the increase in TIMP-2 seen in biopsies taken during a bleeding episode also did not reach statistical significance (P = 0.06) (Fig. 2, C and D). TIMP-3 production was not related to bleeding patterns (Fig. 2, E and F). TIMP-4 was seen in the majority of samples assessed but was unrelated to HT use or to bleeding patterns (P = 0.75).
|
, A and B). TIMP-2 is expressed constitutively and was more widely distributed through the stromal cell population than TIMP-1 and was not commonly present in epithelial cells. TIMP-2 appeared to be localized intracellularly in the majority of stromal cells collected during bleeding episodes (Fig. 2D). TIMP-3 was produced by stromal cells and by glandular and surface epithelium. Although it was strongly expressed in decidualized stromal cells, it was also present at high levels in nondecidualized biopsies from women with irregular bleeding (groups 3 and 4). TIMP-4 was widely distributed through the stromal cell compartment as well as surface and glandular epithelial cells.
Leukocytes
Monoclonal antibodies were employed to identify the main leukocyte subtypes (Table 1 and Fig. 3). The marked increase in CD56+ uNK cells in association with bleeding has already been reported (Fig. 3A) (10). These were the most abundant leukocytes in subjects with abnormal bleeding, and data are included for completeness. The uNK cell concentration was not related to decidualization in this cohort.
|
). CD3+ T cell numbers were lower in HT users. Polymorphic neutrophils were the least prevalent leukocyte population, only three per 1000 stromal cells in HT users with no bleeding and only nine per 1000 stromal cells in those with irregular bleeding (Fig. 3C). CD68+ macrophages were seen in all biopsies, but their numbers varied minimally between the groups (Fig. 3D). Endometrial mast cells (not shown) were identified in less than half of all the biopsies assessed, and because of this variability, the cells were estimated and not counted. Three biopsies from women with irregular bleeding had large populations of mast cells that were degranulated, showing high levels of tryptase activity throughout the extracellular stromal space. CD45RA+ B cells (not shown) were seen in most biopsies in consistently low numbers. It was assumed that they were minor contributors and showed no variability in relation to bleeding episodes.
The endometrial impact of HT exposure was also assessed in longitudinal biopsies obtained before and after HT in the same subject 3 months later (n = 5). Consistent with our previous observations, MMP production did not change with HT exposure. However, TIMP-1 production was markedly increased after HT exposure (Table 4). In paired biopsies, TIMP-1 was further increased in biopsies taken during a bleeding episode compared with biopsies taken from the same subject when she was not bleeding (P = 0.044) (Fig. 3E). No statistically significant changes were seen in TIMP-2, TIMP-3, and TIMP-4 in these paired biopsies (data not shown). Leukocyte numbers in paired biopsies showed no clear relationship to bleeding patterns (data not shown), with the exception of uNK cells as previously reported (10).
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The most striking finding in this study was the increase in TIMP-1 production in association with HT exposure and in relation to irregular bleeding and the marked increase in TIMP-1 seen during bleeding episodes. It is most likely that the increased TIMP-1 seen was a reaction to rather than a cause of bleeding, perhaps reflecting extrauterine delivery from plasma or infiltrating leukocytes in areas of vascular damage or increased permeability.
In our pilot studies comparing TIMP-1 and MMP-9 in postmenopausal women not using HT (controls) with combined HT users, we found low levels of both MMP-9 and TIMP-1 in HT users compared with controls. These studies were conducted using a different technique (in situ hybridization) (28). In the current study, we were most interested in variations in MMP and TIMP production between those HT users with no bleeding and those with irregular bleeding and used labeled antibodies to define MMP and TIMP. Immunoreactive protein was examined in this study to prove that translation of the message occurs. Unfortunately, because of the very limited tissue availability, it was not possible to perform in situ zymography, which would have indicated MMP activity.
In premenopausal endometrium, it is likely that the tissue balance between MMPs and TIMPs regulates endometrial breakdown and repair and that the increased MMP production and activation at the time of menstruation overwhelms the stabilizing effect of TIMPs, promoting vascular breakdown, bleeding, and remodeling (13). In our study, we did not find a statistically significant increase in MMP-3 or -9 in association with bleeding. This could be a function of small sample size but could also suggest that other mechanisms apart from MMP activity regulate abnormal bleeding in HT users.
Proteolytic activity of MMP in the endometrium is partly regulated by TIMPs. TIMP-1 forms a preferential complex with pro-MMP-9, whereas TIMP-2 binds to pro-MMP-2 and facilitates enzyme activation. TIMP-3 binds with both pro-MMP-2 and pro-MMP-9 (29). TIMP-4 is the most recently characterized of the TIMP family. It is a potent inhibitor of MMP-26 and is maximally expressed in the early secretory phase and may be regulated by the estrogen receptor (30). TIMP-4 has not previously been studied in postmenopausal endometrium, but in premenopausal users of the levonorgestrel intrauterine system (Mirena, Schering, Germany), localization of TIMP-4 was not related to bleeding patterns, but biopsies were not obtained during bleeding episodes, as in our study (31).
TIMPs have a number of other important regulatory roles in the endometrium. In addition to the binding and inhibition of active sites on MMPs, TIMPs are also regulated by growth factors and cytokines at the transcriptional level and may inhibit endothelial cell growth, independently of their actions on MMPs (32). We have previously demonstrated that endometrial vascular density is reduced in HT users compared with postmenopausal women not taking HT (9). It has been speculated that abnormal angiogenesis contributes to abnormal bleeding in HT users (33), but there is little evidence to support this. In premenopausal women, abnormal endometrial angiogenesis is thought to contribute to BTB with long-term progestogens (34) and is associated with increased production and activation of MMP (17). Additional studies are required to assess the impact of HT exposure on endometrial angiogenesis.
A role for TIMP-1 in causing irregular bleeding is suggested by our observation that production was increased both during bleeding and in subjects with a recent history of irregular bleeding biopsied outside a bleeding episode.
The aim of this study was to define the role of MMP in irregular bleeding on HT. In keeping with our studies of MMP production during normal menstruation (35) and during abnormal bleeding in users of long-acting progestogen-only contraception (LTPOC) (20), we expected to find increased production of MMP and reduced TIMP in association with irregular bleeding. However, the combination of low MMP production with raised TIMPs seen here is quite different from that seen in LTPOC users where MMP-9 (and leukocytes) were increased and TIMPs 1–3 reduced (18). This is particularly surprising given that depot medroxyprogesterone acetate users are commonly anovulatory and show similar histological appearance to that seen in HT users. This suggests that bleeding on combined HT occurs by a different mechanism than that seen in normal menstruation and with LTPOC. We have recently developed interventions for abnormal bleeding on LTPOC aimed at blocking MMP activity (36 .) The findings from this study would suggest that these interventions may not be effective for irregular bleeding on HT.
Endometrial leukocytes peak premenstrually and are increased in LTPOC users with BTB (37). Leukocytes produce cytokines and MMPs known to break down endometrial vessels and induce bleeding (38). We have previously demonstrated a specific increase in the uNK leukocyte subset in association with bleeding in HT users (10). In this study, we report an overall increase in the endometrial leukocyte population with T cells as the most numerous subgroup, suggesting that leukocytes or their endometrial products may contribute to vascular breakdown in this population. Various types of leukocytes including monocytes (39), lymphocytes (40), and uNK cells (41) express TIMP-1. Hence, it is possible that the increase in TIMP-1-immunolabeled cells observed could be a result of increased endometrial uNK cells seen in association with bleeding. Additional studies are needed to define the role of endometrial leukocytes in postmenopausal endometrium and their relationship to bleeding. This is of clinical significance, because regulation of leukocyte populations may potentially indicate an effective intervention to stop or prevent bleeding.
The majority of biopsies in this cohort were defined as atrophic or very weakly proliferative. Obtaining sufficient endometrial tissue in the outpatient setting from postmenopausal women is problematic, and relatively few hysterectomies are performed in this large regional center on HT users. Only 33% of biopsies taken from groups 1 and 2 contained sufficient tissue for reliable analysis compared with 72 and 78% in groups 3 and 4 with irregular bleeding, raising the possibility that these biopsies may not be representative of this population overall. In premenopausal women using the LTPOC Norplant, abnormal bleeding was associated with low circulating levels of estradiol (42). However, in postmenopausal HT users where endogenous estradiol is extremely low, HT preparations with a higher content of estradiol are associated with increased frequency of abnormal bleeding (43). It is certainly possible that differences in estradiol or progestogens or in the endometrial receptors could have contributed to variations between individuals and groups in this study. Obtaining adequate biopsies from postmenopausal women with no endometrial pathology and a thin endometrium is known to be problematic. In this study, no subject with abnormal bleeding had an endometrial thickness greater than 6 mm, and all subjects with polyps were excluded. Recent studies of outpatient endometrial sampling from this population (44) indicate that there is only a 27% probability of obtaining an adequate endometrial sample for histological analysis from this population, using the Pipelle curette. All biopsies in this study were personally taken by the first author, and we were able to obtain an overall rate of 45% adequate biopsies. Despite the low subject numbers in groups 1 and 2, the tests used still attained more than 80% power to detect differences in MMP production of 50% between the groups for binary outcomes and more than 90% power to detect differences of more than 1.5 SD for continuous outcomes.
The findings from this study do confirm that bleeding in HT users is accompanied by local tissue breakdown. Evidence of tissue breakdown, breeches in vessel walls, interstitial hemorrhages, and epithelial sloughing (24, 25) were seen only in endometrial biopsies taken from subjects with irregular bleeding and those biopsied during a bleeding episode These biopsies were not obtained under direct vision, and it is possible that increased MMP activity occurred locally around bleeding sites and was not detected from blind biopsies. Obtaining directed biopsies from bleeding sites involves the use of hysteroscopy with relatively large-diameter hysteroscopes and requires patient sedation (45). Because hysteroscopy was not clinically indicated in our subjects (endometrial thickness all <6 mm), hysteroscopy using large-diameter hysteroscopes would not have been considered ethically acceptable at our center.
In summary, we have demonstrated that bleeding with combined HT is associated with a significant increase in endometrial TIMP-1 combined with low production of MMP-1, -3, and -14. This suggests that bleeding in HT users occurs in an endometrial environment quite different from that seen in both normal menstrual bleeding and in BTB in progestogen users. Improved understanding of these mechanisms is an important step in the development of effective prevention or treatment strategies.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Author disclosure summary: M.H., J.C., L.A.M., D.A.D., I.S.F., and L.A.S. have nothing to declare.
First Published Online May 9, 2006
Abbreviations: BTB, Breakthrough bleeding; DAB, 3,3'-diaminobenzidine; HT, hormone therapy; LTPOC, long-acting progestogen-only contraception; MMP, matrix metalloproteinase; TIMP, tissue inhibitor of metalloproteinase; uNK, uterine natural killer.
Received December 16, 2005.
Accepted May 3, 2006.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  M. Hickey, D.A. Doherty, I.S. Fraser, D.M. Sloboda, and L.A. Salamonsen Why does menopausal hormone therapy lead to irregular uterine bleeding? Changes to endometrial blood vessels Hum. Reprod., April 1, 2008; 23(4): 912 - 918. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 The ESHRE Capri Workshop Group Endometrial bleeding Hum. Reprod. Update, September 1, 2007; 13(5): 421 - 431. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
, outliers, values exceeding 1.5 x interquartile range; *, extreme values, those outside 3 x interquartile range. E, Graded levels of TIMP-1 immunolocalization in paired biopsies before HT treatment (Gp 1) and subsequent biopsy 3 months later (Gp 2) or paired biopsies between bleeding episodes (Gp 3) and during bleeding episodes (Gp 4).