A conceptual understanding of the mechanisms involved in the onset and cessation of normal menstrual bleeding provides both the foundation and the context for understanding the pathophysiology of anovulatory bleeding.

The classic concepts of normal menstruation derived primarily from direct observations of the cyclic changes in endometrium transplanted from the uterus to the anterior chamber of the eye in nonhuman primates; vascular events played the key role in the explanation for how menses both began and ended.^{31, 32} Basically, menstruation was envisioned as ischemic necrosis of the endometrium caused by vasoconstriction of the spiral arterioles in the basal layer, triggered by withdrawal of estrogen and progesterone. Similarly, the end of menses was explained by longer and more intense waves of vasoconstriction, combined with coagulation mechanisms activated by vascular stasis and endometrial collapse, aided by rapid reepithelialization mediated by estrogen derived from the emerging new follicular cohort.

The results of more contemporary investigations do not support the classic hypoxia theory of menstruation. Perfusion studies in women have failed to demonstrate reduced endometrial blood flow just before menses.³³ Hypoxia-inducible factor (HIF)-1, a nuclear protein that activates gene transcription in response to reduced cellular oxygen (the earliest known marker of response to hypoxia), is barely detectable and not widely distributed in human premenstrual endometrium cultured under hypoxic conditions.³⁴ Histologically, early menstrual endometrium exhibits focal necrosis, inflammation, and coagulation rather than the diffuse hyalinization or coagulation necrosis that would be expected to result from vasoconstriction and hypoxia.³⁵ Slowly but surely over the last decade or so, the operational paradigm for menstruation has shifted. Instead of vascular events, the central theme of the new model of the initiation of menstruation is an enzymatic autodigestion of the functional layer of the endometrium and its subsurface capillary plexus, possibly extending to the spiral arteriolar system in the basal layer.³⁵ The classic concept of the mechanisms that end normal menstruation is essentially unchanged; coagulation mechanisms, local vasoconstriction, and re-epithelialization all contribute to hemostasis in the menstrual endometrium with vascular events playing the key role.

The enzymatic degradation of the endometrium triggered by estrogen-progesterone withdrawal involves a number of different but interrelated mechanisms including the release of intracellular lysosomal enzymes, proteases from infiltrating inflammatory cells, and the actions of matrix metalloproteinases. In the first half of the secretory phase, acid phosphatase and other potent lytic enzymes are confined to intracellular lysosomes, their release inhibited by progesterone via stabilization of lysosomal membranes. As estrogen and progesterone levels fall in the days preceding menses, lysosomal membranes destabilize and the enzymes within are released into the cytoplasm of epithelial, stromal, and endothelial cells, and eventually, into the intercellular space. These proteolytic enzymes digest their cellular constraints as well as surface membranes and desmosomes (intercellular bridges). In the vascular endothelium, their actions result in platelet deposition, prostaglandin release, vascular thrombosis, extravasation of red blood cells, and tissue necrosis.^{35, 36}

Progesterone withdrawal also stimulates an inflammatory response in the endometrium. Just before menstruation, the total number of leukocytes in the endometrium increases
markedly to as much as 40% of the stroma.^{37, 38} The inflammatory infiltrate (including neutrophils, eosinophils, and macrophages or monocytes) is drawn by chemo-attractive molecules (chemokines) synthesized by endometrial cells, some of which are downregulated by progesterone (interleukin 8; IL-8).³⁷ When activated, the leukocytes produce a wide assortment of regulatory molecules including cytokines, chemokines, and a range of enzymes that contribute to degradation of the extracellular matrix, directly, or indirectly, via activation of other proteases.

Matrix metalloproteinases are a family a proteolytic enzymes that degrade components of the extracellular matrix and basement membrane.³⁹ The metalloproteinases include collagenases that degrade interstitial and basement membrane collagens, gelatinases that further digest collagens, and stromelysins that attack fibronectin, laminin, and glycoproteins. Each member of the family is substrate specific and secreted as an inactive zymogen that requires activation by plasmin, leukocyte proteases, or other metalloproteinases. The expression, secretion, and activation of endometrial matrix metalloproteinases is cycle dependent and increases markedly in the late secretory phase just before menstruation.^{40, 41} Overall, progesterone inhibits endometrial metalloproteinase expression, an action mediated by transforming growth factor (TFG)-b.⁴² Progesterone withdrawal has the opposite effect—increased metalloproteinase secretion and activation, followed by dissolution of the extracellular matrix.⁴³ Local modulators (predominantly cytokines), derived from endometrial epithelial, stromal, and endothelial cells, and natural tissue inhibitors of matrix metalloproteinases that bind the active form of the enzymes also play an important role in their regulation.⁴⁴ In cycles of conception wherein elevated progesterone levels are sustained, matrix metalloproteinase activity remains effectively suppressed. In the normal menstrual cycle, metalloproteinase expression is suppressed again after menses, presumably by increasing estrogen levels.

Progressive enzymatic degradation of the endometrium eventually disrupts the subsurface capillary and venous vascular system, causing interstitial hemorrhage; dissolution of the surface membrane allows blood to escape into the endometrial cavity. Ultimately, degeneration extends to the deepest extent of the functional layer where rupture of the basal arterioles contributes to bleeding. A natural cleavage plane develops at the junction of the loose, vascular, edematous stroma with the basal layer. Desquamation begins in the fundus and gradually extends toward the isthmus. The end result is the typical deflated and shallow but dense menstrual endometrium.^{45, 46}

The menstrual fluid is comprised of an autolysed endometrium rich in inflammatory exudates, red blood cells, and proteolytic enzymes.^{35, 46} One of those enzymes, plasmin, formed by activation of its inactive precursor, plasminogen, has potent fibrinolytic actions that help to prevent clotting of menstrual fluid and to facilitate the expulsion of degenerated tissue. Plasminogen activators that mediate the conversion of plasminogen to plasmin are found in late secretory and menstrual endometrium and are released from degenerated endometrial vascular endothelium.^{35, 46} The volume of menstrual bleeding is controlled, at least to some extent, by the local balance between fibrinolysis and clotting. Endometrial stromal cell tissue factor and plasminogen activator inhibitor (PAI)-1 promote clotting and help to balance fibrinolytic processes.^47,48 and ⁴⁹ Early in menstruation, intravascular platelet plugs, and later, thrombi form at the shedding surface, helping to limit blood loss. Their importance to hemostasis in the menstrual endometrium can be inferred from the increased volumes of menstrual blood loss observed in women with thrombocytopenia and von Willebrand disease. Ultimately, however, the cessation of menstrual bleeding depends on vasoconstriction in the denuded spiral arterioles in the basal layer of the endometrium, and also possibly in the radial arteries of the superficial myometrium. Endothelins are potent long-acting vasoconstrictors of vascular smooth muscle produced by endometrial glandular, stromal, and endothelial cells. Menstrual endometrium contains high concentrations of endothelins and prostaglandins, which together cause intense vasoconstriction in the spiral arterioles.³⁵
The myometrial contractions associated with menstrual events very likely reflect the actions of prostaglandin F_2a, but in contrast to postpartum bleeding, myometrial contractions are not important for control of menstrual bleeding.

Surface re-epithelialization also contributes to hemostasis in the menstrual endometrium. The process occurs very rapidly, beginning at the mouth of the basal portions of residual glands in areas otherwise completely denuded, and spreading outward. The peripheral regions of the cavity at the isthmus and near the tubal ostia (which do not shed during menses) also contribute to its resurfacing.^{38, 46} Generally, by cycle day 5, these scattered areas of epithelial proliferation converge and fuse; bleeding stops completely only when the new epithelial surface is complete. The mechanisms that govern this initial phase of tissue repair and the role that estrogen has, if any, are uncertain. In the first few days of the new cycle, circulating estrogen levels and endometrial estrogen and progesterone receptor concentrations are low and unchanged from premenstrual levels.^{50, 51} Moreover, even after oophorectomy and vigorous endometrial denudation, the endometrium heals, suggesting that the initial phase of tissue repair is largely independent of estrogen.^{38, 46}

The stroma regenerates from stem cells located in the basal layer of the endometrium, but only after a confluent surface epithelium has been restored. Damaged endometrial vessels are quickly repaired. New vessel growth and mitotic activity in all parts of the regenerated human endometrium coincide with increasing serum estrogen levels and rising endometrial estrogen and progesterone receptor concentrations.^{38, 46} Matrix metalloproteinases present in the menstrual endometrium and other proteases may be important mediators of the release and activation of growth factors needed for endometrial repair. Vascular endothelial growth factor is an important promoter of endometrial mitosis and can be induced by tumor necrosis factor (TNF)-a, TGF-b, and insulin-like growth factor-1.^{35, 52, 53} Experimental evidence derived from model systems suggests that activins and other members of the TGF-b superfamily also may play a role.^{38, 54}

One clinical example of estrogen withdrawal bleeding is that which may follow bilateral oophorectomy during the follicular phase of the cycle. The bleeding that occurs after removal of the ovaries can be delayed by exogenous estrogen therapy, but will occur when treatment stops. Other examples include cyclic estrogen-only hormone therapy in castrate or postmenopausal women and the midcycle bleeding that can accompany the transient but abrupt fall in estrogen levels immediately preceding ovulation.

The best clinical examples of estrogen breakthrough bleeding are the different patterns of bleeding observed in women with chronic anovulation. The amount and duration of estrogen breakthrough bleeding can vary widely, depending on the amount and duration of unopposed estrogen stimulation that the endometrium has received. Relatively low levels of chronic estrogen exposure typically result in intermittent spotting or staining that is generally light in volume but may be prolonged. In contrast, sustained high level estrogen stimulation commonly results in long intervals of amenorrhea punctuated by acute episodes of often profuse bleeding that vary in duration.

Progestogen withdrawal bleeding is observed when treatment with exogenous progesterone or a synthetic progestin is discontinued. Progestogen withdrawal bleeding usually occurs only when the endometrium has first been primed with endogenous or exogenous estrogen. The amount and duration of bleeding can vary widely and generally correlates with the level and duration of previous estrogen-stimulated endometrial proliferation. In women with marginal to frankly low estrogen levels or short intervals of amenorrhea, bleeding is generally light to scant and may not occur at all. In those with sustained high estrogen levels or long intervals of amenorrhea, bleeding can be heavy and somewhat prolonged, but still is self-limited. Between the extremes, the amount and duration of bleeding induced by progestogen withdrawal is typically similar to that observed at the end of a normal ovulatory cycle. In women receiving cyclic hormone therapy with exogenous estrogen and progestin, bleeding follows withdrawal of progestin even if estrogen treatment continues; progestin withdrawal bleeding can be delayed, but only if estrogen levels are increased by 10-20 fold.⁵⁵

Progestogen breakthrough bleeding occurs when the ratio of progestogen to estrogen is unfavorably high. Unless there is sufficient estrogen to balance its action, continuous treatment with exogenous progesterone or synthetic progestins will result in intermittent bleeding of varying duration that is generally light, a pattern very similar to low level estrogen breakthrough bleeding described above. Clinical examples of progestogen breakthrough bleeding are the bleeding observed in women using the progestin-only contraceptive “minipill” or other long-acting progestin-only contraceptive methods (progestin implants, depot medroxyprogesterone acetate).⁵⁶ The breakthrough bleeding observed in women using combination estrogen-progestin contraceptives also is a form of progestogen breakthrough bleeding. Although all estrogen-progestin contraceptive regimens contain pharmacologic quantities of both estrogen and progestin, the progestin component is always the dominant hormone and the net effect on the endometrium is profoundly progestational.

Anovulatory bleeding can result from estrogen withdrawal bleeding, reflecting the transient fall in estrogen levels that accompanies regression of a follicular cohort, or from estrogen breakthrough bleeding, due to focal break down of an overgrown and structurally fragile endometrium under continuous estrogen stimulation. The heaviest episodes of anovulatory bleeding tend to occur in women with sustained high levels of estrogen; women with polycystic ovary syndrome, obese women, postmenarcheal adolescents, and perimenopausal women are common clinical examples. The clinical presentation spans the spectrum from the pale frightened teenager who has bled for weeks to the older woman who is deeply concerned that she may have cancer.

In contrast to the organized predictable pattern of sequential estrogen-progesterone stimulation and withdrawal that characterizes the normal ovulatory menstrual cycle, the patterns of ovarian steroid hormone production and endometrial stimulation in anovulatory women are disorganized and unpredictable. By definition, the anovulatory woman is always in the follicular phase of the ovarian cycle and in the proliferative phase of the endometrial cycle. There is no luteal or secretory phase because there is no ovulation or cycle. The only ovarian steroid signal the endometrium receives is estrogen, levels of which constantly fluctuate, rising and falling as each new cohort of follicles begins to grow but ultimately loses its developmental momentum and, sooner or later, lapses into atresia. Although the amplitude of the signal may vary, the message, growth, stays the same.

Over a period of time, an unrelenting, uninterrupted estrogen growth stimulus can stimulate the endometrium to proliferate to abnormal heights where it becomes fragile. Without the growth limiting and organizing effects of progesterone, the endometrium lacks the stromal support structure to maintain stability. Focal areas breakdown and bleed and, as those areas heal under the influence of continued estrogen stimulation, others break down and bleed. Persistent proliferative and hyperplastic endometrium characteristically exhibits numerous discrete foci of stromal breakdown near the epithelial surface, associated with pools of extravasated red blood cells, capillary platelet/fibrin thrombi, and repair-related changes recognized as ball-like aggregates of tightly packed stromal cells beneath a cap of intact but hypertrophied epithelium.³⁵ The cause for the focal breakdowns in persistent proliferative endometrium is not entirely clear. However, abnormal endometrial growth involves not only epithelial and stromal cells but also the microvasculature.

Venous capillaries in persistent proliferative and hyperplastic endometrium are increased, dilated, and often form abnormal irregular channels; ultrastructural studies have revealed a number of abnormal structural elements that predispose to fragility.^{57, 58} The abnormal microvasculature could be the result, but is more likely the proximate cause, of abnormal bleeding. The weight of available evidence from histologic and molecular studies indicates that anovulatory bleeding results from an increased density of abnormal vessels having a fragile structure prone to focal rupture, followed by release of lysosomal proteolytic enzymes from surrounding epithelial and stromal cells and migratory leukocytes and macrophages. Once initiated, the process is further aggravated by local release of prostaglandins, with greater sensitivity to those that vasodilate (PGE₂) than to those that vasoconstrict (PGF_2a).⁵⁹ Other molecules (perforins) inhibit capillary plug formation and further degrade the capillary venous network. Vasoconstriction of basal endometrial and superficial myometrial vessels does not occur because tissue loss is only focal and superficial, and does not typically reach the basal layer where denudation triggers an intense vasoconstrictive response. The final mechanism that normally controls menstrual bleeding, surface epithelial reconstruction, operates in persistent proliferative endometrium, but not in a normal way. Epithelial repair is focal, in the areas of breakdown, not universal; the result is a constantly changing patchwork of small repairs instead of an organized and well structured remodeling.³⁵

Laboratory tests can be very helpful but are not always necessary. A sensitive urine or serum pregnancy test can quickly exclude the possibility that abnormal bleeding relates to a complication of pregnancy; a positive test leaves only a few diagnostic possibilities, which generally are not difficult to distinguish. A complete blood count to exclude anemia and thrombocytopenia is prudent in all women with complaints of abnormal bleeding, especially when heavy or prolonged.

After excluding pregnancy, the most important question to answer is whether the patient is ovulating, because the causes and clinical management of ovulatory and anovulatory uterine bleeding are quite different. When the menstrual history alone does not allow a confident conclusion, a well timed serum progesterone determination during the putative luteal phase of the cycle can help to document ovulation or anovulation. A logical strategy is to obtain the test between cycle day 22 and 24, after ovulation in the longest normal cycle and before the end of the shortest normal cycle; any value greater than 3 ng/mL provides reliable evidence that ovulation has occurred recently.⁸⁵ However, when bleeding episodes are frequent or poorly documented, proper timing for a progesterone measurement can be difficult to determine. It also is important to remember that many women with abnormal bleeding, especially perimenarcheal and perimenopausal women, ovulate at least occasionally. Although most commonly applied to assess ovulatory function in women with infertility and now seldom used even for that purpose, basal body temperature recordings can be very informative in women with a confusing pattern of bleeding. Endometrial biopsy also can be used to assess ovulatory function (proliferative vs. secretory endometrium) but cannot be justified for that purpose alone when a less costly and less invasive serum progesterone measurement provides the same qualitative information; endometrial sampling should be reserved for those at risk for endometrial hyperplasia or neoplasia, as discussed below.

In sexually active women, a nucleic acid based test for chlamydia and gonorrhea and a wet prep to exclude trichomonas infection merit consideration, particularly in those with evidence of vaginitis and/or cervicitis. In presumed or proven anovulatory women, a serum thyroid-stimulating hormone (TSH) level excludes any associated thyroid disorder. Liver or renal function tests are indicated only for those with known or strongly suspected disease.

Adolescents, women with a suspicious personal or family history of bleeding symptoms (easy bruising, frequent gum bleeding when flossing or brushing teeth, epistaxis), and women with unexplained menorrhagia warrant evaluation with coagulation studies to exclude coagulopathies, such as von Willebrand disease, factor deficiencies, and platelet function abnormalities.^{66, 68, 86, 87} In addition to a platelet count, screening should include
both a prothrombin (PT), which evaluates the extrinsic and final common clotting pathways, and an activated partial thromboplastin time (aPTT), which tests the intrinsic and common pathways of coagulation. Although the PT and aPTT have relatively low positive and negative predictive value for detecting underlying bleeding disorders,⁸⁸ they are adequate screens for severe factor deficiencies.⁸⁹ The high prevalence of von Willebrand disease among women with menorrhagia (approximately 13%) warrants specific exclusion of the diagnosis and justifies measurement of von Willebrand factor, ristocetin cofactor activity (von Willebrand factor activity), the factor VIII level, and blood typing.^{87, 90, 91} It is important to note that test results can fluctuate over time,⁹² and also may vary across the menstrual cycle; repeated testing, ideally during the first few days of the cycle, may be required to establish the diagnosis of von Willebrand disease.^90,93 The blood type is helpful because von Willebrand factor and factor VIII levels are 25% lower in patients with type O blood than in those with other blood types.⁹⁴ Although the bleeding time is the traditional method for evaluating platelet function, an automated laboratory test (Platelet Function Analyzer, PFA-100) is taking its place because it has greater sensitivity and reproducibility and is less invasive.^{95, 96} The instrument exposes platelets in citrated whole blood to high shear inside a capillary tube and monitors the drop in flow rate as the platelets form a plug in the center of a membrane coated with collagen and either adenosine diphosphate or epinephrine. For patients with abnormal coagulation studies, consultation with a hematologist is recommended.^{97, 98}

An endometrial biopsy can exclude endometrial hyperplasia or cancer. Age over 35 or 40 years is widely considered a risk factor for endometrial disease and cited as an indication for biopsy in women with abnormal bleeding. Endometrial hyperplasia and cancer are more commonly detected in older than in younger women, but the duration of exposure to unopposed estrogen stimulation is the more critical risk factor. Longterm exposure is more likely in older than in younger women, but women under age 30, and even teenagers, can develop endometrial cancer.^99,100,101 and ¹⁰² In premenopausal women, the likelihood of abnormal endometrial histology is relatively high (14%) when menses are irregular, but very low (< 1%) when cycles are regular.¹⁰³ The small flexible suction cannulas now widely available cause less discomfort than older traditional biopsy instruments and yield comparable results.^104,105 and ¹⁰⁶ Unfortunately, hospital-based curettage without hysteroscopy is still commonly performed, even though it is no longer the gold standard.

In addition to revealing any intrinsic endometrial disease, such as chronic endometritis, hyperplasia, or adenocarcinoma, biopsy can help to direct further evaluation or to guide the choice of treatment in women with a confusing history of abnormal bleeding. An inactive or atrophic endometrium identifies women unlikely to respond to progestational therapy. In women with no recent exposure to exogenous progestins, a secretory endometrium provides reliable evidence of recent ovulation and signals the need to search for an anatomical cause.

Imaging can help to differentiate anovulatory bleeding from anatomical causes, myomas and endometrial polyps being the most common examples. Standard transvaginal ultrasonography can provide accurate information about the size and location of any uterine fibroids that may explain abnormal bleeding or exaggerate the bleeding due to other causes.¹⁰⁷

Ultrasonography may reveal an obvious cavitary lesion or an abnormally thin or thick endometrium. A very thin endometrial “stripe” (<5 mm), like a biopsy that yields minimal tissue, suggests an attenuated or denuded endometrium best treated first with estrogen rather than with a progestin or an estrogen-progestin combination (discussed below).

In perimenopausal and postmenopausal women with abnormal bleeding, endometrial biopsy generally is considered unnecessary when the endometrial thickness is less than 4 or 5 mm because the risk of endometrial hyperplasia or cancer is remote.^108,109 and ¹¹⁰ It seems logical to apply the same criterion for the same reason in premenopausal women with abnormal bleeding, although there is no substantial direct evidence to support the extrapolation. Otherwise, the decision to biopsy or not should be based primarily on clinical suspicion and risk factors rather than on ultrasonographic measurements of endometrial thickness. That does not mean that endometrial thickness has no bearing on the decision whether to perform a biopsy; a grossly increased endometrial thickness (>12 mm) increases the risk of disease and is an indication for sampling, even when clinical suspicion of pathology is otherwise low.¹¹¹ In summary, we believe that biopsy is unnecessary when the endometrial thickness is less than 5 mm, that biopsy is indicated when the clinical history suggests long-term unopposed estrogen exposure even when the endometrial thickness is “normal” (5-12 mm), and that biopsy should be performed when endometrial thickness is greater than 12 mm even when clinical suspicion of disease is low.