Danforth’s Obstetrics and Gynecology

D. Ware Branch and James R. Scott

RECURRENT MISCARRIAGE

Recurrent miscarriage (RM), traditionally defined as three or more consecutive first-trimester spontaneous losses, affects up to 1% of couples. Primary recurrent miscarriage is diagnosed in women who have never had a successful pregnancy, and secondary recurrent miscarriage in those whose repetitive losses follow a live birth. There is no specific classification for women who have multiple miscarriages interspersed with normal pregnancies. It is generally agreed that a workup for possible causes of RM is indicated in most patients after two or three consecutive miscarriages.

The management of couples with RM is controversial. This clinical entity has received much attention in the lay and medical literature during the past decade. A definite cause is established in no more than 50% of couples, and several alleged causes of RM are controversial. Despite publicity to the contrary, there is little evidence that poor nutrition, infections, unrecognized diabetes, toxic agents, or psychological trauma are significant etiologic factors. Some alleged experts and Internet sites inappropriately emphasize unproven hypotheses and results from poorly designed clinical studies. Seeking a solution, some patients and physicians may explore less well-accepted etiologies and empirical or alternative treatments. Moreover, new diagnostic tests for RM are continually being proposed to replace those that have been disproved and discarded. For example, antithyroid antibodies, elevated follicular-phase luteinizing hormone levels, circulating maternal embryotoxic factor, and abnormal lymphocyte subset ratios (elevated CD56 + levels) have been touted within the last decade. It is beyond the scope of this chapter to critically analyze each new “treatment” or assay, but the mechanism of pregnancy loss and potential relationship to each of these remains largely theoretical. Until effective treatments are identified and proven by properly designed studies, these screening tests have little use in the routine evaluation of patients with RM.

Typically, investigation of anatomic, hormonal, genetic, and infectious factors has been recommended. Even these may be criticized because the derivation of their diagnostic use and treatments advocated are empirical and have come under scrutiny because they were never submitted to properly designed study. Importantly, evidence has mounted that the average women with RM has a fairly good prognosis for a successful next pregnancy without any specific treatment.

Known and Suspected Causes of Recurrent Miscarriage

Structural Uterine Defects Hysterosalpingography, magnetic resonance imaging, hysteroscopy, sonohysteroscopy, and laparoscopy can be used to diagnose septate uterus, other müllerian anomalies, uterine defects associated with diethylstilbestrol exposure, submucous myomas, and intrauterine synechiae. The prognosis for successful pregnancies in patients with müllerian anomalies is related to the type of malformation, with asymmetric fusion defects carrying the worst prognosis and septate, bicornuate, and didelphic uteri carrying increasingly better prognoses. In patients with RM, the prevalence of these anatomic defects is approximately 10% to 15%. The cause of pregnancy loss in women with uterine anomalies is uncertain. A diminished blood supply interfering with normal implantation and placentation and the reduced size of the uterine cavity are often cited as possible causes, but these reasons seem unlikely or insubstantial in the setting of the arcuate uterus. Abdominal metroplasty has been replaced in most cases by the hysteroscopic removal of uterine septa. This procedure can be accomplished in an outpatient setting and eliminates the need for cesarean delivery. Noncontrolled, retrospective studies suggest the subsequent live-birth rate is greater than 80%. Removal of synechiae and submucous myomas can also be performed hysteroscopically.

Endocrine Problems The luteal phase defect (LPD) has long been thought to be a cause of spontaneous abortion, but the evidence linking LPD to recurrent abortion is subject to criticism. It is traditionally thought that women with LPD have short menstrual cycles, postovulatory intervals less than 14 days, and secondary infertility. LPD was initially thought to be due to failure of the corpus luteum to make enough progesterone to establish a mature endometrial lining suitable for placentation. This theory has evolved to implicate poor follicular-phase oocyte development, which results in disordered estrogen secretion, inadequate ovarian steroidogenesis, and subsequent maldevelopment of endometrial receptors. In turn, these effects could result from excess luteinizing hormone or hyperandrogenic states. Some investigators claim that LPD accounts for over one fourth of cases of RM, but studies of this disorder have not included concurrently tested controls. Also, there is little agreement on the criteria necessary to make the diagnosis. Endometrial biopsy or luteal-phase serum progesterone levels are the most widely accepted diagnostic tests. Both are timed for the late luteal phase of the cycle. The endometrial biopsy is histologically dated, and a lag greater than 2 to 3 days is considered suspect. However, this should be confirmed by repeat biopsy, because delayed endometrial histology can occur sporadically in women with no reproductive problems. To further confuse the issue, normal women have endometrial histology suggestive of LPD in up to 50% of single menstrual cycles and 25% of sequential cycles. Though the association between LPD and RM remains speculative, many clinicians have treated women with RM with progesterone in their next pregnancy. One commonly advocated treatment is a 25-mg progesterone suppository inserted into the vagina twice a day (morning and night) beginning after ovulation and continuing until menses begin or through the first 8 to 10 weeks of pregnancy. Comparable doses of oral micronized progesterone have also been used. No properly designed studies have evaluated the role of progesterone treatment in women with RM with LPD. Older studies and meta-analyses are difficult to interpret because of marked differences in inclusion criteria and how LPD was diagnosed, the use of various progesterone compounds, and the small number of women studied. In a more recent randomized trial, a subgroup of women with polycystic ovary syndrome (PCOS) and three or more miscarriages were randomized to treatment with either progesterone or placebo pessaries. There was no difference in the pregnancy outcomes. Clomiphene and other ovulatory agents have been tried to improve follicular development and corpus luteum function, but the results have been variable. Human chorionic gonadotropin has been used in an attempt to stimulate the corpus luteum support of pregnancy in women with RM. One placebo-controlled, multicentered trial found no significant difference in the successful pregnancy rates (83% vs. 79%). In summary, the relationship between the LPD and recurrent pregnancy loss remains a subject of controversy. It has not been shown conclusively that progesterone treatment or corpus luteum support influences pregnancy outcome in women with recurrent pregnancy loss. PCOS has been found in one third or more of women with RM. However, the diagnosis of PCOS in women with RM does not predict a worse pregnancy outcome than in women with RM without PCOS. There is no known effective therapy for women with PCOS and RM.

Genetic Abnormalities Parental chromosomal anomalies are found in approximately 3% to 5% of couples with RM. Cytogenetic examination of both partners is helpful to predict recurrence and forms the basis for genetic counseling. Most abnormalities are balanced translocations, with two-thirds being reciprocal translocations and one-third robertsonian translocations. Couples with balanced translocations have spontaneous loss rates ranging from 50% for reciprocal translocations to 25% for robertsonian translocations. All couples with a parental chromosomal abnormality deserve counseling about genetic amniocentesis or chorionic villus sampling in any future pregnancy to exclude a serious fetal chromosomal abnormality. Parental chromosomal abnormalities do not usually preclude further attempts at pregnancy, because most couples eventually have normal offspring. For the rare homologous robertsonian translocation that prevents successful pregnancy, therapeutic possibilities include artificial donor insemination, in vitro fertilization with donor oocytes, and adoption. Chromosomal analysis of the products of conception is also clinically useful, particularly in the evaluation of the reason for failure of a treatment regimen. Molecular mutations that may be shown in the future to cause recurrent miscarriages include lethal, single-point mutations, possibly linked to MHC genes; mutations in genes that code for products critical for normal development; mutations in homeobox genes that control transcriptional regulation; mutations that lead to severe metabolic errors and embryonic death; and disorders of protooncogenes and oncogenes. One group has shown that certain polymorphisms of the HLA-G gene are associated with significantly higher rates of miscarriage among couples presenting with RM. Also, marked skewing of the normal 50:50 distribution of X chromosome inactivation in the mother, a condition termed highly skewed X-chromosome inactivation, may be associated with otherwise unexplained RM. Before testing is recommended, confirmation of these molecular genetic associations in different populations is required. For now, commercially available tests for these conditions are not widely available, and there are no proven treatment options.

Autoimmune Disorders

Antiphospholipid syndrome (APS) has been recognized as a proven cause of pregnancy loss for over a decade. Approximately 5% to 15% of women with RM have lupus anticoagulant (LA), anticardiolipin (aCL), or both. These acquired antiphospholipid autoantibodies are induced by as yet unknown stimuli in the setting of aberrant immunoregulation. Low levels of immune globulin G or immune globulin M aCL are of questionable significance. Though women with APS may present with RM in the first trimester, fetal death in the second or early third trimesters may be more specific for the condition. Patients with high levels of aCL or a history of prior fetal death are at greatest risk of another fetal loss. The cause of fetal death appears to be a decidual vasculopathy that results in decidual infarction and insufficient blood flow to the placenta. Intervillous thrombosis has also been described. However, these lesions are nonspecific, and the degree of pathology is not always sufficient to explain the fetal death. The mechanisms by which aCL may cause decidual vasculopathy and fetal death are unknown. A number of pathophysiologic mechanisms have been proposed, including an imbalance of local prostacyclin and thromboxane production, enhanced platelet aggregation, decreased activation of protein C, increased tissue factor, and decreased trophoblast annexin V production or availability. Most recently, the complement system has been invoked as having a major role in antiphospholipid syndrome-related pregnancy loss. Maternally administered heparin is widely considered the treatment of choice for APS pregnancies, both to improve embryo-fetal outcome and protect the mother from thrombotic. Treatment is usually initiated in the early first trimester after ultrasonographic demonstration of a live embryo. The dose of heparin required for safe and effective treatment, however, is debated. Some experts use relatively low doses of heparin (e.g., 5000 U of standard heparin b.i.d.), particularly when treating women with recurrent preembryonic or embryonic losses. However, higher doses of heparin are recommended for patients with APS with prior thrombosis, and some experts urge full anticoagulation. The optimal dose of heparin is controversial for women whose APS is diagnosed because of prior fetal loss or neonatal death after delivery before 34 weeks gestation due to severe preeclampsia or placental insufficiency, but who do not have a history of thromboembolism. These women are at risk for thromboembolic disease, and it is our opinion that these cases should receive sufficient thromboprophylaxis. Low molecular-weight heparins (LMWHs) are widely used in Europe for the treatment of APS pregnancy, and there is little reason to suspect that the appropriate use of LMWHs differs from that of standard heparin with regard to efficacy. In most case series and trials, daily low-dose aspirin is included in the treatment regimen. One important caveat deserves mention—a small, placebo-controlled trial found that otherwise healthy women with RM and low titers of antiphospholipid antibodies do not require treatment.

Intravenous immune globulin has also been used during pregnancy, usually in conjunction with heparin and low-dose aspirin, especially in women with particularly poor past histories or recurrent pregnancy loss during heparin treatment. However, a randomized, controlled, pilot study of intravenous immune globulin treatment during pregnancy in unselected APS cases proved negative. Anticoagulant coverage of the postpartum period in women with APS and prior thrombosis is critical. We prefer switching the patient to warfarin thromboprophylaxis as soon as she is clinically stable from delivery. In most cases, an international normalized ratio of 3.0 is desirable, and postpartum coverage should extend for 6 to 8 weeks after delivery. Because of their risk for thrombosis, the same strategy is recommended in women without prior thrombosis but in whom APS is diagnosed because of prior fetal loss or neonatal death after delivery at or before 34 weeks gestation for severe preeclampsia or placental insufficiency. Both heparin and warfarin are safe for nursing mothers. The need for postpartum anticoagulation in women with primary APS diagnosed solely on the basis of recurrent preembryonic and embryonic losses is unclear.

Other Autoimmune Disorders Autoantibodies to thyroid antigens are associated with a modest increased rate of pregnancy loss if identified in early pregnancy or immediately before pregnancy. Some investigators have found a significant proportion of women with RM to have antithyroid antibodies; others have not. Even if antithyroid antibodies are associated with RM, no treatment options have proven beneficial. Approximately 15% of women with RM have detectable antinuclear antibodies (ANA). Subsequent pregnancy outcomes among women with a positive ANA test result are similar to those among women with a negative ANA test result. A randomized treatment trial of women with recurrent pregnancy loss and a positive autoantibody result, including ANA, found no benefit to treatment with prednisone and low-dose aspirin and treatment with placebo. Thus, currently available data do not support testing women with recurrent pregnancy loss for ANA.

Thrombophilic Disorders The relationship between inherited thrombophilic disorders and recurrent miscarriage has been the subject of intense study within the last several years. The most common inherited thrombophilic disorders are factor V Leiden and prothrombin G20210A mutation, found in approximately 8% and 3%, respectively, of Caucasian women in the United States. These mutations are associated with approximately 25% of isolated thrombotic events and approximately 50% of familial thrombosis. Other less common thrombophilias include deficiencies of the anticoagulants protein C, protein S, and antithrombin III. Hyperhomocysteinemia, most commonly due to the C677T polymorphism of the methylenetetrahydrofolate reductase (MTHFR) gene, is also associated with venous thrombosis. Data regarding the association of these thrombophilic abnormalities and RM do not allow clear and consistent conclusions. The rather obvious fact that most women with common thrombophilic mutations, such as factor V Leiden, the prothrombin G20210A mutation, or the MTHFR C677T mutation, do not have RM further confounds the picture. One prospective study of next pregnancies in women with RM found a significantly lower successful pregnancy rate among those with the factor V Leiden mutation compared to those without (37.5% vs. 69.3%). Various studies are more consistent in finding an association between thrombophilias and second- or third-trimester fetal loss. The odds ratio for stillbirth is significantly higher in women with combined thrombophilic defects. Some women with RM have evidence of ongoing, perhaps chronic, thrombin generation or the formation of thrombosis-related microparticles. These studies underscore the potential importance of prothrombotic states to pregnancy loss, but more research is required to bring the current findings into the clinical realm. Despite the recent interest in this field, no treatment trials have been performed. Thus, which therapy, if any, is effective in promoting successful pregnancy among women with recurrent pregnancy loss and thrombophilia is uncertain.

Cervical Incompetence Incompetent cervix, also called premature cervical dilation, is an important cause of second-trimester pregnancy loss. It is characterized by gradual, painless dilation of the cervix with bulging and rupture of the membranes and subsequent expulsion of a fetus too immature to survive. Pregnancy loss from this cervical abnormality usually occurs in the second trimester and is thought to be an entirely different and distinct entity from a first-trimester miscarriage or premature labor in the third trimester. It results from different factors, presents a distinctive clinical picture, and requires different management. Moreover, miscarriage and premature labor are common, but mid-trimester premature cervical dilation is relatively rare. Unlike the rest of the uterus, the cervix is fundamentally a connective tissue structure. The cause of cervical incompetence is obscure, and various etiologic factors have been proposed. Previous surgery or trauma to the cervix, such as D&C, amputation, conization, cauterization, loop electrosurgical excision procedure (LEEP), or traumatic delivery, seem to be factors in some cases. In other instances, congenital cervical structural defects, uterine anomalies, or abnormal cervical development associated with in utero diethylstilbestrol exposure appear to play a role. Little agreement can be found regarding the diagnosis of cervical incompetence, except that it is one of exclusion that requires careful evaluation to rule out other potential causes of mid-trimester pregnancy loss. Other causes of very early delivery include abruptio placentae, chorioamnionitis, and uterine anomalies, but they usually present different clinical pictures. Whether or not the condition can be diagnosed during the nonpregnant state by methods designed to calibrate the diameter of the endocervical canal or during early pregnancy by sonographic findings is questionable. The absolute diagnosis of cervical incompetence can be made only by seeing the fetal membranes bulging through the partially dilated cervix of a patient in the second trimester of pregnancy who is not in labor. More typically, a presumptive diagnosis is made from the characteristic history of apparently silent dilation of the cervix followed by rupture of the membranes and a relatively painless, rapid labor with delivery of an immature infant. Also, the fetus is typically alive at the time of presentation to the hospital; delivery of a dead, macerated fetus makes the diagnosis of cervical competence questionable. Upon inspection in the nonpregnant state, the cervix may be shortened with a patulous os or may be deformed with lacerations that sometimes extend to the vaginal fornix. Although bed rest, various intravaginal devices, and pharmacologic agents have been used with some success, the generally accepted treatment for incompetent cervix is surgical. Various methods have been described, but the McDonald or Shirodkar procedures are most commonly employed prophylactically. These are techniques performed vaginally, usually under regional anesthesia, designed to reinforce the cervix close to the level of the internal os. If there is insufficient cervical tissue to allow placement of a cerclage vaginally, an abdominal approach is sometimes used. The reinforcement suture is usually placed toward the end of the first trimester after ultrasound documentation of a live fetus, after the risk of miscarriage has passed, and before the cervix starts to dilate.

Placement of the cerclage in the second trimester after cervical change has occurred is sometimes necessary but appears less effective. The procedure should not be used if the diagnosis is in doubt, if membranes are ruptured, or if vaginal bleeding and cramping are part of the clinical picture. There is no evidence that postoperative antibiotics, progesterone, or tocolytic agents are useful adjuvants. If membranes rupture or labor ensues at any time, removal of the cerclage should be strongly considered to prevent chorioamnionitis, sepsis, cervical laceration, and rupture of the uterus. Otherwise, the suture is removed when fetal maturity is achieved, usually after 37 weeks gestation, which is often followed by the onset of labor and a relatively rapid delivery. If the patient desires further pregnancies, some physicians leave the cerclage in place and deliver by cesarean section. The procedure has also been used prophylactically for patients with previous preterm deliveries with less convincing evidence for cervical incompetence. Results from prospective randomized studies have led to conflicting conclusions.

Idiopathic Causes Because most cases of RM have no discernible cause, alloimmune factors have long been suspected. These have yet to be proven, largely because little is known about the mechanisms that prevent immunologic rejection of the conceptus in successful pregnancies. Early reports proposed that HLA compatibility between couples, the absence of maternal leukocytotoxic antibodies, or the absence of maternal blocking antibodies were related to RM. The importance of these factors has not been substantiated, and these expensive tests are no longer clinically indicated. Contemporary research is focused on local decidual or trophoblast immunosuppressive factors such as cytokines, growth factors, hormones, enzymes, and endometrial proteins. Some of these immunoactive factors appear to be necessary for implantation and growth and development of the early placenta and embryo, and others may cause abortion, when expressed. There are, however, no practical clinical tests available for these factors and no proven treatment if they were found abnormal. Although no alloimmune mechanism has been unequivocally shown to cause RM in humans, several types of immunotherapy have been advocated. Originally, the attempt to improve maternal immunotolerance in recurrent aborters was based on evidence that pretransplant blood transfusions decreased rejection of organ allografts and that the rate of resorptions or abortions in animal models was reduced by prior immunization with spleen cells from a paternally related strain. The most popular regimen involves injections of the father's leukocytes. Though proponents persist, this treatment is questionable at best and harmful at worst. Most randomized trials have proven negative, and the largest and only multicenter randomized trial found that treated pregnancy outcomes were worse in the women who received leukocyte immunization. Based largely on this trial, the U.S. Food and Drug Administration has stated that the administration of this therapy for RM may only be done as part of a clinical investigation, and then only if there is an investigational new drug application in effect. Participating women should be counseled that immunization using viable leukocytes carries the risks of any blood transfusion, such as hepatitis, human immunodeficiency virus, and cytomegalovirus infections. Reactions have been uncommon but include soreness and redness at the injection site, cutaneous graft-versus-hostlike reaction, fever, maternal platelet and leukocyte alloimmunization, and blood group sensitization. Intravenous immune globulin has been proposed as an alternative therapy in patients with idiopathic RM. A number of randomized trials have been reported, and the results are conflicting. Nevertheless, this treatment seems to be no more successful than paternal cell immunization, and intravenous immune globulin is not recommended outside of a research protocol by either the American College of Obstetricians and Gynecologists or the American Society of Reproductive Medicine. It is imperative for physicians to recognize that the prognosis for idiopathic RM is by no means dismal. Numerous studies and several meta-analyses indicate that the average next pregnancy live-birth rate for placebo-treated women with idiopathic RM is 60% to 70%. Many couples see this modestly favorable prognosis in a somewhat positive light, and it compares favorably with the prognosis for conditions such as APS or parental karyotype abnormalities. Understanding this prognosis may allow the couple to choose against an expensive unproven treatment. One caveat—as expected, increasing maternal age and increasing number of miscarriages are negative variables.

Recommendations for Recurrent Miscarriage

A sympathetic attitude by the physician is crucial—establishment of trust and rapport and a sincere appreciation of the distress and grief experienced by these couples permit tactful and thorough discussions with patient and partner. It is reasonable to institute an evaluation after two consecutive miscarriages in anxious women or if the patient has few reproductive years remaining or has had an infertility problem. Couples interested in an investigational protocol are perhaps best referred to legitimate research centers.