Please cite this paper as: Garry R, Hart R, Karthigasu KA, Burke C. Structural changes in endometrial basal glands during menstruation. BJOG 2010;117:1175–1185.
Objective To prospectively observe the changes occurring in endometrial glandular morphology during menstrual shedding and regeneration.
Design Prospective observational study.
Setting The academic gynaecological endoscopy unit of a university teaching hospital.
Population Thirteen patients investigated for a variety of benign, non-infective gynaecological disorders during the active bleeding phase of the menstrual cycle.
Methods The morphological appearances of concurrent histological and scanning electron microscopic images of endometrium taken at different stages of the active bleeding phase of menstruation were studied and correlated with the simultaneous immunohistochemical expression of the Ki–67 proliferation marker and the CD68 marker of macrophage activity.
Main outcome measure Change in morphology of endometrial glands at various stages of menstruation.
Results Endometrial glands within the basalis show evidence of apoptosis and associated macrophage activity immediately before and during menstruation. There is subsequent destruction and removal of old secretory glandular epithelial elements, and rapid replacement with new narrow glands lined with small epithelial cells. There is no evidence of mitotic cell division or expression of Ki–67 in the glandular cells during this replacement process, but there is evidence of marked macrophage activity prior to glandular cell loss.
Conclusions Early endometrial epithelial repair after menstruation is not a consequence of mitotic cell division. It occurs without evidence of Ki–67 expression. There is structural evidence of programmed cell death and intense macrophage activity associated with glandular remodelling. Dead epithelial cells are shed from the glands and accumulate within the endometrial cavity to be replaced by new small epithelial cells that appear to arise by differentiation of the surrounding stromal cells. We propose that these stromal cells are endometrial progenitor/stem cells.
We have recently reported on the morphological changes in the surface of exposed endometrium after menstrual shedding.1 These data suggested that new surface endometrial epithelial cells arise from underlying stromal cells, rather than as outgrowths from glandular epithelial cells. This current study extends these observations to the concomitant sequential morphological changes observed in the deep basalis glands throughout the menstrual phase of the cycle.
The study was approved by the King Edward Memorial Hospital’s ethics committee. Endometrium was obtained from 34 patients with benign gynaecological conditions undergoing routine gynaecological assessment. Of these, 13 women were between day 28 and day 3 of their menstrual cycle, and required hysterectomy4 or endometrial curettage.9 This group was studied most intensively for this project. Histories were obtained prospectively and accurate dating of the stage of the cycle and time of onset of menstruation was prospectively recorded. Of the women in the perimenstrual stage of the cycle, two were at day 28 of regular cycles, and were judged from their symptoms to be immediately premenstrual. Three were within the first 24 hours of menstruation, and five were in day 2 and three in day 3 of the bleeding phase of the cycle.
Obtaining endometrium from the menstrual phase of the cycle presents a number of specific practical and social difficulties, and is rarely available in routine clinical practice. Five women in this study had endometriosis, two had adenomyosis without endometriosis, four had menstrual disturbances and two presented with fertility problems. It is recognised that these pathologies may influence the appearances and mechanisms of menstruation, but endometrial sampling and/or hysterectomy is not performed without medical indications, and this represented the only material available to us.
The study was of a prospective observational design, primarily of women in the menstrual phase of the cycle. Numerous blocks were obtained from each sample of endometrium. These were processed to show conventional histological, immunohistochemical and scanning electron microscopic appearances of the same endometrial specimens.
Haematoxylin and eosin (H&E)
Biopsies were fixed in 10% formalin for 18 hours and embedded in paraffin. All biopsies had a standard H&E stain and were submitted to an experienced histopathologist for identification of any pathological features.
A standard indirect immunoperoxidase procedure using mouse anti-rat Ab, biotinylated sheep anti-mouse and streptavidin-horseradish peroxidise (S–HRP), as previously described,2 was used. Whole-mount sections were incubated in a blocking solution for 1 hour at room temperature (18–22 c), followed by primary antibody to CD34 (DakoCytomation, Glostrup, Denmark) at a concentration of 0.1 μg/ml (clone QB End 10),9 antibody to CD68 (Sapphire Bioscience Pty Ltd, Redfern, NSW, Australia), antibody to c–Kit (DakoCytomation, Glostrup, Denmark) or antibody to Ki–67 (Invitrogen, Mount Waverley, Victoria, Australia Pty), with incubation times of 30–45 minutes at room temperature.
Negative controls were included in each run, substituting the primary antibody for an iso-type (IgG1,kappa) non-specific immunoglobulin at the same concentration (0.1 μg/ml) as the primary antibody. Immunopositive cells were counted by a single masked observer, and expressed as an overall proportion of immunopositive cells to the total number of cells in the sample.
Scanning electron microscopy (SEM)
Squares of endometrium with sides of about 10 mm were pinned out using fine needles with the luminal surface uppermost. They were immersed in 2.5% gluteraldehyde in 0.01 M phosphate-buffered saline, pH 7.4, for 18 hours at 4°C. The tissues were then dehydrated and critically point dried. The dried samples were coated with both carbon and gold and viewed using a Zeiss Supra variable pressure FESEM or a Philips 505 SEM. Images were captured digitally as TIFF files.
Normal pre- and postmenstrual endometrial appearance
We first demonstrate the universally accepted characteristic appearance of late secretory phase endometrium. The glands are almost invariably vertically orientated and at right angles to the luminal epithelium. They are of relatively large diameter with substantial lumens and irregular outer borders (Figure 1A, B). These surface and glandular epithelia are made up of characteristic tall columnar epithelium, with many microvilli, interspersed with numerous ciliated cells (Figure 1C).
The equally well recognised classic appearance of proliferative phase endometrium is of numerous glands of smaller diameter than those observed during the secretory phase, many of which run predominantly horizontally and parallel with the luminal surface of the endometrium for large parts of their length (Figure 1D, E). Glands and surface epithelium at this early phase of the cycle are made up of smaller cuboidal cells with a flatter surface and fewer microvilli (Figure 1F).
Early menstrual appearance
In the late secretory phase (Figure 2A) and early menstrual phase (Figure 2B), collections of macrophages expressing CD68 antigen are observed in and around the basal glands. This appearance is observed principally between days 28 and 2 of the cycle. The macrophages observed are larger and more obvious than at other stages of the cycle (Figure 3). They may be observed infiltrating (black arrows) between columnar epithelial cells of the glands (Figure 2C), and can even be found within the lumen (white arrows) of the glands at this stage of the cycle (Figure 2D). There is often also histological evidence of nuclear fragmentation (black arrows), which is suggestive of apotosis and programmed cell death associated with these macrophage appearances during the earliest phases of menstruation (Figure 2E, F). The proportion of CD68+ macrophages is also demonstrated in a formal cell count (Table 3). This demonstrates that macrophages are most commonly found in and around the glands on day 28 and during the phase of active menstrual bleeding.
Glandular shedding and regrowth
Menstruation is characterised by the piecemeal loss of the outer functionalis layer of the endometrium, leaving areas of exposed basalis stroma through which stumps of the glands initially protrude. Immediately after the loss of the superficial layers of the endometrium, stumps of glands (white arrows) and vessels (black arrows) come to lie horizontally on the surface of the exposed basalis stroma (Figure 4A). Far from appearing the healthy source of new epithelial cells, these most superficial residual glands show features of glandular disintegration, with openings within the usually closed cylindrical gland (Figure 4A). In the earliest stages of menstrual shedding there is a striking difference in the morphology between some of the more superficial, obviously late secretory phase endometrial glands and other glands that are narrow and straight, and lined by low, cuboidal epithelial cells, suggestive of new early proliferative phase glands (Figure 4B). Close-up views of an SEM of a residual gland stump protruding above the basalis at this stage shows evidence of multiple deficiencies in the glandular wall that again suggests the disintegration rather than proliferation of these exposed glands (Figure 4C). A similar appearance with partial loss of the circumference of some of the glands is seen on a standard H&E image of an early day–1 endometrium (Figure 4D). During the earliest stages of menstrual bleeding, individual columnar cells may be observed desquamating from the gland wall (Figures 4E), with associated cellular debris collecting within the glandular lumen (Figures 4E, F). Concurrent with this phase of desquamation, a smaller type of flat cell is observed lining some areas of these glands, which are now mostly horizontally aligned within the basalis (Figure 4F). These probable ‘new’ glands are characterised by the findings of small cuboidal cells lining some or the entire circumference of the gland, combined with much cellular debris within the gland lumen (Figure 4G). It is unlikely that these combined changes of epithelial shedding and glandular cellular debris are artefacts or the result of trauma to the specimens, for in some areas a clear stepwise and progressive increase in the size of the cells is observed within a single gland (Figures 4G). In a day–1 sample, in which areas of unshed secretory phase endometrium are located nearby, there is evidence of areas that are already re-epithelised with new cuboidal epithelium. In this sample there are several very thin-walled blood vessels running horizontally and parallel with the luminal surface. There is what appears to be a collapsed secretory phase gland running in the same horizontal direction, beneath which is an entirely different, thin gland of typical proliferative phase morphology lined by cuboidal epithelium (Figure 4H).
Evidence of proliferation during the menstrual cycle
Ki–67 is a well-recognised proliferation marker that is extensively expressed in the endometrium. The extent of its expression is dependent on the phase of the cycle. This antigen is very sensitive to the conditions used in the preparation of the specimens. All specimens illustrated here and all calculations were obtained from the same immunohistochemical processing run, and were obtained under the same conditions.
The expression of Ki–67 is maximal during the estrogen-dominated first half of the cycle, when it can be seen in all elements of the endometrium, including the surface and glandular epithelium (white arrows), vascular endothelial cells (black arrows) and many cells in the stroma (Figure 5A). Ki–67 is also expressed during the progesterone-dominated secretory phase of the cycle, but at this stage, cells demonstrating this antigen are largely confined to stroma (black arrows), with almost no mitotic activity in the surface or glandular cells (Figure 5B).
Of particular note is that during days 1–3 of menstruation, when much of the regeneration of surface and glandular epithelium occurs, there is very little evidence of cellular proliferation in any cellular element of the endometrium, and virtual none in the residual glandular elements (Figure 5C). Figure 5D shows a day–1 endometrium with the very early development of new surface epithelium associated with glands showing a variety of different morphologies. There are intact glands lined by columnar epithelial cells, glands lined by partially shedding columnar epithelium, glands lined by small ‘new’ epithelial cells and gland cavities containing much cellular debris. None of the glands show evidence of Ki–67 expression. New surface epithelium appears first as small, flattened cells on a fibrinous matrix (Figure 5D, E). Some residual glands join up with new surface epithelium, but this process again occurs without evidence of gland cell proliferation (Figure 5F, G), and the new surface epithelial cells are cuboidal, consisting mainly of spherical nuclei, and without obvious microvilli (Figure 5H).
This study is an attempt to determine the dynamics involved in the evolution of a late secretory phase into early proliferative phase endometrium (Figure 1A–D). How do wide, convoluted, vertically-orientated secretory phase glands become straight, narrow, often horizontally orientated, early proliferative phase glands? How do highly differentiated tall columnar surface and glandular secretory phase epithelial cells evolve to become small, less differentiated cuboidal cells?
The current theory of early endometrial regeneration
The late secretory phase endometrium contains many, vertically aligned, wide-diameter glands lined by tall columnar epithelium (Figure 6A). During menstruation the outer functionalis layer of the endometrium is shed, removing most of the functionalis. Stumps of glands are retained within the basalis, and these have been suggested to be the major source of the new surface epithelial glands (Figure 6B). Novak and Te Linde, in their landmark paper on the histology of the menstruating uterus, observed that:
The source of new epithelium is chiefly the epithelium of the basal stumps of the uterine glands. Strange to say, mitosis is not a very frequent occurrence, appearing to a greater extent when the epithelial layer is already complete.3
Subsequently Ferenczy,4 on the basis of pioneering SEM studies, observed that the ‘surface epithelium is derived from proliferation from the exposed ends of basal glands’, while also noting in the same paper that there is a ‘lack of estrogen-dependent morphogenic alterations (mitosis and ciliogenesis) in the uterine mucosa during the regenerative period’. This theory may be illustrated by the residual glandular epithelial cells undergoing mitotic cell division, and then spreading by amoeboid-type migration across the denuded surface of the basalis stroma (Figure 6C). This concept remains currently the most widely accepted explanation for the mechanisms of early endometrial repair.5–7.
Objection to the current theory of endometrial regeneration: Ki–67 expression
In the proliferative phase of the cycle many cells in all compartments, including surface and glandular epithelial cells, express the proliferation marker Ki–67, and can be observed to be actively undergoing mitotic cell division (Figure 5A). As the secretory phase progresses there is progressively less evidence of cell division within epithelial cells, but there continues to be many actively dividing stromal cells. Many of these dividing stromal cells are uterine natural killer (uNK) cells.8 We have previously shown that menstrual shedding is a piecemeal process, and that the regeneration of new epithelium occurs concomitantly with functionalis loss.1,9 During this phase of rapid early surface epithelial regeneration there is virtually no expression of Ki–67 in either the old or the new epithelial cells (Figures 5C and 7).
If Novak and Te Linde’s widely accepted theory for endometrial regeneration were to be correct, we would expect to find evidence of cellular health and mitotic activity in the glandular stumps within the basalis. We in fact found unhealthy glandular epithelium, with evidence of cellular degeneration and cell death. In addition to the lack of mitotic activity, we have observed three sets of findings that suggest that residual basal glands are not the primary source of new surface and glandular epithelium.
The morphological features of cells undergoing apoptosis include cell shrinkage, reorganisation of the nucleus, membrane, nuclear blebbing and finally the fragmentation of the nucleus into apoptotic bodies.10 We observed these features in some day 1–2 glands. Such physical changes of programmed cell death have been previously documented in proliferative phase endometrium,11 but the process is relatively infrequent at this stage of the cycle, and is known to be more common in the late secretory and premenstrual phases,12–15 and to peak on the second menstrual day.16 The association between apoptosis and tissue breakdown that occurs at the time of menstruation has led some investigators to postulate a mechanistic role of apoptosis in the process of tissue breakdown.17 In tissue culture experiments, however, dissociation between tissue breakdown and apoptosis was demonstrated, suggesting that although both processes can occur at the same time, they are not mechanistically related.18 It seems probable therefore that apoptosis alone is not responsible for the cellular renewal that occurs within the basalis during menstruation.
Macrophages and apoptotic cells are positively correlated in the endometrium,19 and macrophages are known to be involved in the clearance of apoptotic cells.20 They have been detected in the endometrium throughout the cycle, and increase in numbers from the proliferative through to the menstrual phase.21 On days 27–28 they make up 6–15% of total endometrial cells,22 and there are reports of them concentrating around endometrial glands.23 Our work confirms that macrophages concentrate around (Figure 2A, B), and indeed invaginate between (Figure 2C, D), glandular epithelial cells, particularly in the glands within the basalis on day 28 and days 1–2 of menstruation Figure 3. They can also be found within the cavity of the glands (Figure 2D). The macrophages are larger and more polymorphic in the perimenstrual phase compared with their morphology at other stages of the cycle. It is therefore possible that macrophages and programmed apoptotic cell death are involved in the process of endometrial gland replacement and renewal during menstruation. We, however, suggest that these mechanisms alone are insufficient to account for the very rapid changes in cellular and glandular morphology that are observed during menstruation.
Epithelial cell shedding
We have observed, on the basis of SEM and histological images, that glands remaining in the basalis after shedding of the functionalis during menstruation show partial loss of the integrity of the gland lining (Figure 4A–D). During days 1–2 of menstruation, individual glandular epithelial cells may be observed to separate from the surrounding gland framework (Figure 4E), with the associated accumulation of shed epithelial cells within the lumen of the cavity. During the earliest stages of menstrual shedding some glands show large areas of cell debris accumulation, associated with the partial or complete lining of the now narrow glands with very small cells (Figure 4F).
We postulate that Figure 4G indicates the nature and the speed of the regeneration process. This endometrium is taken from a woman who had begun to bleed 12 hours earlier. The functional endometrium is completely shed, and a fibrinous mesh is forming on the denuded surface. There is cellular debris in the lumen of the gland. There is an area on the surface of the gland nearest the uterine cavity that is devoid of classical glandular epithelial cells, and is lined only by a few very small cells consisting almost entirely of an elongated narrow nucleus. Either side of these cells are a few cells of intermediate size with rounder nuclei, and these merge with normal looking columnar cells lining the remainder of the circumference of the gland. Our hypothesis is that the secretory cells lining this gland had been shed into the lumen of the gland, as evidenced by the cellular debris, and replaced by new cells arising from very small undifferentiated cells placed near the surface of luminal and glandular damage that very rapidly differentiate into regular glandular epithelial cells (Figure 6D).
This cellular differentiation theory of the origin of new endometrial epithelium is in disagreement with the currently widely accepted theory of cellular division, but it is not original. In 1933, the great pathologist George Papanicolaou observed in a guinea-pig model that:
The rapidity of new epithelisation and lack of mitotic figures cannot be well explained by the generally accepted theory that the uterine epithelium regenerates from the epithelium of the deep glands which have escaped destruction.24
He further observed that:
The glands are desquamated to the very tip of the gland and then a process of regeneration of an entirely new glandular epithelium quickly ensues. The changes within the glands did not appear to be synchronous with changes on the surface of the uterus, and this may have been the reason why they escaped the attention of previous investigators.24
He described a stage of desquamation:
…during which the epithelium disintegrates and falls into the lumen of the gland which is soon filled with a large number of desquamated cells and leucocytes. The next stage is characterized by the gradual differentiation of the superficial cells of the tunica propria. These cells are small and have round or slightly elongated, oval, or elipitical nuclei and a very small amount of cytoplasm. They resemble undifferentiated embryonic cells and are very abundant within the tunica propria…These cells become rounded and gradually form a continuous lining on the surface of the naked gland.24
The observations by Papanicolaou were made only on guinea-pigs. We believe our paper is the first description of the same phenomenon in humans. We have quoted the 1933 paper extensively because the description almost exactly mirrors our observations 76 years later. The concept of epithelial renewal by differentiation from stromal cells has also been supported by Baggish et al.25, who observed that ‘the residual glandular epithelium was metabolically inactive and unlikely to be the source of young growing cells’. They further suggested that rather a group of stromal cells lying in the stromo–epithelial border might play an active role in the regeneration of the surface epithelium by a metaplastic process.
Our new theory of early endometrial regeneration
In addition to apoptotic and macrophage-related glandular remodelling of basal glands during menstruation, there is also shedding of columnar glandular epithelial cells into the lumen of these glands (Figure 5D). In some areas very small, viable cells are seen lining parts of these de-epithelised glands. There is a rapid and progressive increase in size of the glandular cells from very small flattened cells through to ‘mature’ cuboidal and then columnar epithelial cells (Figure 4G).
These observations of rapid structural changes that appear to occur within hours or even minutes of glandular cell loss, have led us to a novel cellular-differentiation hypothesis of endometrial regeneration after menstruation. We postulate that the cells lining the new narrow glandular structures arise not as a result of mitotic cell division but of differentiation of adjacent endometrial stromal cells. Such amitotic-produced new glands tend to be arranged as long narrow structures in horizontal planes parallel with the uterine cavity lumen, and following the course of the many new thin-walled blood vessels that arise within the remodelled endometrium.
We believe that new glands, surface epithelium and probably new blood vessels arise in the endometrium during menstruation, without evidence of cell division. We suggest that the most likely alternative explanation for this phenomenon is that the new epithelial cells appear as a consequence of cell differentiation from progenitor or stem cells within the endometrium. Recent brilliant work from the Melbourne group of Gargett26 has demonstrated that the endometrium contains cells that exhibit both extensive clonal self-renewal properties and the multipotent capacity to differentiate into cytokeratin+ glands, as well as smooth muscle cells, adipocytes, chondrocytes and osteoblasts. Clonality and multipotent capacity to differentiate are the primary characteristic functional properties of progenitor and/or stem cells. The endometrium is now known to contain functionally active stem cells, and it is likely that this type of cell is responsible for the amitotic regeneration of endometrial glandular and surface epithelium during endometrial repair after menstrual shedding.
Early endometrial epithelial repair after menstruation is not a consequence of mitotic cell division. It occurs without evidence of Ki–67 expression. There is structural evidence of programmed cell death, and intense macrophage activity associated with glandular remodelling. Dead epithelial cells are shed from the glands and accumulate within the endometrial cavity, to be replaced by new small epithelial cells that appear to arise by differentiation of the surrounding stromal cells. We propose that these stromal cells are endometrial progenitor/stem cells.
Disclosure of interests
There are no conflicts of interest with any of the authors and the contents of this report.
Contribution to authorship
RG conceived the study, collected, correlated and analysed the data, undertook a number of the procedures and prepared the manuscript. RH, KAK and CB collected the clinical cases, were responsible for some of the procedures, and commented on and revised the manuscript.
Details of ethics approval
The trial was considered and approved by the Ethics Committee of King Edward Memorial Hospital, Subiaco, Perth, WA 6008, Australia.
We are grateful for a research grant from AGES (The Australian Gynaecological Endoscopic Society).
We acknowledge the excellent technical and laboratory support of the scientists of the Department of Obstetrics and Gynaecology, University of Western Australia, and for their help in the preparation and interpretation of the histological and SEM images.
1. Background: Discuss existing theories for endometrial regeneration.
2. Methods: This study reports a small select sample of women who attended for hysterectomy or curettage as treatment for benign disorders. Discuss the implications for external validity.
Antigen Ki-67 was used in this study as a proliferation marker, but it is very sensitive to laboratory conditions. Discuss the implications.
3. Results and Implications: Describe and explain Figure 5A and Figure 6D. In Figure 7, contrast the percentage of cells expressing Ki-67 in the menstrual-early repair (1) and proliferative (2) phases.
One of the findings in this study that contradicts older theories is that of unhealthy, as opposed to regenerative, epithelium in the glandular stumps of the basalis after menstruation. Critically appraise the possible impact of the selected sample on such findings.
4. Future research: In view of the limitations of this study but its novel findings and important conclusions, discuss with your peers suitable future research to confirm or refute the findings.
Were the findings of this study to be confirmed with future research, what would be the implications for the management of menstrual disorders? What would be the implications for theories about the origin of endometriosis?