Subinvolution of the Placental Site as an Anatomic Cause of Postpartum Uterine Bleeding: A Review

Postpartum uterine bleeding is defined as uterine blood loss greater than 500 mL in the case of a vaginal delivery and greater than 1000 mL after cesarean birth. Primary postpartum bleeding occurs within the first 24 hours after completion of the third stage of labor, while secondary (delayed) postpartum bleeding occurs after this time frame.1 The risk factors and associated uterine pathology of primary postpartum bleeding are well known and include uterine atony, retained placenta, inherited disorders of coagulation, consumptive coagulopathy, endometrial infection, and perineal lacerations. Uterine atony is regarded as the most common physiologic abnormality responsible for primary postpartum bleeding; therefore, obstetricians are well informed with regard to the diagnosis and treatment of uterine atony.2,3 Secondary postpartum bleeding, in contrast, is less frequently addressed; however, it is estimated to complicate about 1% of all pregnancies. Although primary and secondary postpartum bleeding may share common causes, the etiology of secondary postpartum bleeding often remains unknown if the patient can be managed conservatively.4 If endometrial curettage is performed, surgical pathologists examine uterine curettings in order to establish an etiology of the bleeding. Occasionally, pathologists are confronted with a postpartum hysterectomy specimen in severe cases. However, excluding the presence of chorionic villi, endometritis, and gestational trophoblastic disease in this clinical setting is insufficient evaluation. Though possibly underrecognized by general surgical pathologists, subinvolution of the placental site (placental site vascular subinvolution or subinvolution of uteroplacental arteries) is an important contributor to secondary postpartum bleeding. Subinvolution of the placental site refers to delayed or inadequate physiologic closure and sloughing of the superficial modified spiral arteries at the placental attachment site. Subinvolution can be identified and documented by the typical clinical features and the histologic findings in postpartum endometrial curettings and hysterectomy specimens. Although detailed studies have been performed by several authors to determine the normal and abnormal anatomy and histology of the uterine placental site,5–12 there is a relative dearth of practice-oriented literature on the topic of subinvolution. This article briefly reviews the physiology of normal uteroplacental arterial development and involution and the proposed pathophysiology of subinvolution. Comparisons to another disease of the uteroplacental vasculature, preeclampsia, are also considered. Using case material from our own institution and descriptions offered in the literature, the essential histologic findings of subinvolution are illustrated. Establishing a diagnosis of subinvolution is critical, as this process implies an idiopathic, and not iatrogenic, cause of delayed postpartum uterine bleeding.

The developing fetus ultimately derives its nutrition from the maternal blood supplied to the placenta via the maternal uterine spiral arteries. Early in gestation (6–10 wk), placenta-derived extravillous cytotrophoblasts (EVTs) invade the decidual portion of the maternal spiral arteries, initially forming endovascular “plugs” that actually impede blood flow to the developing placenta.7 As gestation moves into the second trimester, the endovascular EVTs replace the normal endothelial lining of the artery and express endothelial-type markers and angiogenic factors.13,14 The EVTs also take residence within and around the walls of the spiral arteries (interstitial EVTs). This vascular colonization process extends as deep as the inner third of the myometrium.5–7 The interstitial EVTs work to remodel the arterial walls, and a histological correlate of this remodeling is the observation of hyaline fibrinoid material that replaces the arterial media, with loss of elastic fibers.12 Trophoblast-independent mechanisms of vascular alteration are also at work during this time but are beyond the scope of this discussion. The net physiologic effect of this remodeling and colonization process is the creation of large-caliber, high-flow, but low-resistance arteries sufficient to meet the demands of the growing placenta and fetus.

After delivery, a physiologic mechanism of uteroplacental arterial involution is required to eliminate these altered vessels. In the third trimester, this process begins modestly as the endovascular EVTs are replaced by maternal-derived endothelial cells. The exact temporal relationships are not fully elucidated, but several involution events occur within the uteroplacental vascular bed in the few days after delivery. These changes include occlusive fibrointimal thickening, endarteritis, thrombosis, regeneration of the internal elastic lamina, and disappearance of both endovascular and interstitial EVTs (Figure 1, A and B). Although the histomorphology of involutional events is well described, the exact molecular basis that triggers these changes remains poorly understood.15–17 Necrosis and sloughing of the decidua and superficial endomyometrium occur in tandem with the involutional vascular changes. Contraction of the uterine smooth muscle also contributes to mechanical shrinkage and involution of the placental site as a whole.9 Taken together, one very important consequence of these changes is to limit the loss of blood at the placental site after delivery.

The EVTs obviously play an important role in the development of adequate uteroplacental arterial flow, and an abnormal interaction between these cells and the maternal tissue is proposed as the mechanism for subinvolution.16 An immunologic basis for this type of altered fetomaternal interaction has been suggested and is an especially attractive idea when compared to the findings in preeclampsia. In preeclampsia, poor investment and remodeling of the spiral arteries by EVTs is found. Mounting epidemiologic and molecular evidence suggests that poor interaction between EVTs and maternal decidual tissue may be due to an abnormal immunologic recognition process.8,18–20 The respective pathophysiology of preeclampsia and subinvolution may represent opposite ends of a spectrum of abnormal cellular interactions at the fetomaternal tissue interface. Other evidence supporting an abnormal immunologic basis for subinvolution includes the observations by Andrew et al15 of differential immunoglobulin and complement deposition in involuted and subinvoluted vessels. In this study, the deposition of immunoglobulins (IgG, IgA, IgM) and complement proteins (C1q, C3d, C4) in the walls of normally involuted vessels is described; these components are absent in subinvoluted vessels. Another insight into the mechanism of subinvolution is related to EVT apoptosis. While the EVTs in preeclampsia appear to lose expression of the antiapoptosis protein Bcl-2, increased expression of this protein has been reported in subinvolution.20,21 Perhaps persistent expression of this antiapoptosis protein plays a role in maintaining the uteroplacental vessels in a state similar to gestation in cases of subinvolution. Despite these interesting observations, there remains no uniform consensus as to a defined pathophysiologic sequence of events in subinvolution.

The diagnosis of subinvolution can be made by pathologic examination of postpartum endometrial curettage material, provided the placental site is sampled. Vascular subinvolution may be detected in the presence of rare degenerating chorionic villi, which is a valuable observation when the amount of villous tissue present does not account for bleeding. Therefore, the diagnosis of subinvolution should be carefully excluded in all cases, and the temptation to only record the presence or absence of villous tissue should be avoided.

A number of microscopic features have been described in the literature.11,12,15–17 The essential findings identifiable on hematoxylin-eosin staining include large, patent, and dilated superficial myometrial vessels with pink hyaline material replacing the media and partial intravascular thrombosis of variably aged thrombi (Figure 2, A and B). These abnormal vessels are typically clustered with little intervening myometrial tissue and may be located adjacent to normally involuted vessels (Figure 2, C). Extravillous interstitial and/or endovascular trophoblasts should also be identified. The cytologic features of EVT cells include polygonal shape, abundant amphophilic cytoplasm, and vesicular nuclei (Figure 2, D). Immunohistochemistry may be helpful to identify these cells, as they stain readily with commercially available antibodies directed at low-molecular-weight cytokeratin, inhibin-α, Mel-CAM, and human placental lactogen (Figure 3, A and B).22 Rare cells may also stain with β human chorionic gonadotrophin. Other, though less constant, microscopic features have been described in subinvolution. These include lack of internal elastic lamina duplication, partial or complete absence of a true endothelial lining, and immunohistochemical expression of the antiapoptosis protein Bcl-2. Histochemical stains for elastin (elastic von Gieson or elastochrome) can be very helpful in delineating the vessel wall anatomy. Immunoperoxidase staining for vascular endothelial markers may also be helpful in defining the extent of endothelial regeneration, particularly anti-CD31. The constellation of histologic findings in cases of subinvolution likely reflects the nature of this process as a biologic continuum of changes, lying somewhere between appropriate involution and complete lack of it. The extent of subinvolution varies from case to case, and the diagnostic findings may be quantitatively limited in any given specimen. In cases where the complete set of essential findings is not observed, subinvolution may be suggested as a diagnostic possibility. A diagnosis of subinvolution should only be rendered in the appropriate clinical setting, which is secondary (delayed) postpartum bleeding. Physiologic patency of uteroplacental arteries will be seen in the immediate (<24 h) postpartum period and should not be unequivocally interpreted as subinvolution.

Delayed postpartum bleeding should always raise the possibility of subinvolution. Although subinvolution may cause bleeding anytime between 1 week and several months postpartum, the most common reported timeline for presentation is within the second week after delivery.16 In our case files, we identified 8 patients since 1997 with a pathologic diagnosis of subinvolution (3 hysterectomies, 5 curettings). As can be seen in the Table, 3 of our patients presented within the second week postpartum, 3 presented between the third and seventh weeks postpartum, and time of presentation was not available for the remaining 2 patients. Subinvolution has also been reported to cause bleeding after induced abortion23 and after molar pregnancy evacuation.24 The severity of the bleeding is also variable, but typically there is an abrupt onset of increased uterine bleeding, which prompts the patient to seek medical attention. Endometrial curettage may not be curative, and continued brisk bleeding may require hysterectomy in spite of fluid and coagulation maintenance. Clinical associations cited in the literature include multiparity and older age.16 Four of the patients in our case file were multiparous, and the age range was 19 to 34 years, but 7 of the 8 patients were 27 years of age or older.

Subinvolution of the placental site is a cause of delayed postpartum bleeding. This process is likely underrecognized by general surgical pathologists. The histologic findings of dilated, clustered, myometrial arteries partially occluded by thrombi of various ages and invested with extravillous trophoblasts are diagnostic of subinvolution in the appropriate clinical setting. The cause of subinvolution is not known, but this process may be a manifestation of an abnormal interaction between fetal-derived trophoblasts and maternal tissue. Subinvolution is an important process to recognize, as it implies an idiopathic, and not an iatrogenic, cause of delayed postpartum bleeding.