CHAPTER 37: Gestational Trophoblastic Disease

Gestational trophoblastic disease (GTD) refers to a spectrum of interrelated but histologically distinct tumors originating from the placenta (Table 37-1). These diseases are characterized by a reliable tumor marker, which is the β-subunit of human chorionic gonadotropin (β-hCG), and have varied tendencies for local invasion and spread.

Gestational trophoblastic neoplasia (GTN) refers to the subset of GTD that develops malignant sequelae. These tumors require formal staging and typically respond favorably to chemotherapy. Most commonly, GTN develops after a molar pregnancy but may follow any gestation. The prognosis for most GTN cases is excellent, and patients are routinely cured, even with widespread metastases. The outlook for preservation of fertility and for successful subsequent pregnancy outcomes is equally bright (Vargas, 2014; Wong, 2014). Accordingly, although GTD is uncommon, because the opportunity for cure is great, clinicians should be familiar with its presentation, diagnosis, and management.

The incidence of GTD has remained fairly constant at approximately 1 to 2 per 1000 deliveries in North America and Europe (Drake, 2006; Loukovaara, 2005; Lybol, 2011). Although historically higher incidence rates have been reported in parts of Asia, some of this disparity may reflect discrepancies between population-based and hospital-based data collection (Chong, 1999; Kim, 2004; Matsui, 2003). Improved socioeconomic conditions and dietary changes may be partly responsible as well. That said, certain Southeast Asian populations as well as Hispanics and Native Americans living in the United States do have increased incidences (Drake, 2006; Smith, 2003; Tham, 2003).

Maternal age at the upper and lower extremes carries a higher risk of GTD (Altman, 2008; Loukovaara, 2005). This association is much greater for complete moles, whereas the risk of partial molar pregnancy varies relatively little with age. Moreover, compared with the risk in those aged 15 years or younger, the degree of risk is much greater for women 45 years (1 percent) or older (17 percent at age 50) (Savage, 2010; Sebire, 2002a). One explanation relates to ova from older women having higher rates of abnormal fertilization. Similarly, older paternal age has been associated with increased risk (La Vecchia, 1984; Parazzini, 1986).

A history of prior unsuccessful pregnancies also increases the risk of GTD. For example, previous spontaneous abortion at least doubles the risk of molar pregnancy (Parazzini, 1991). More significantly, a personal history of GTD increases the risk of developing a molar gestation in a subsequent pregnancy at least 10-fold. The frequency in a subsequent conception is approximately 1 percent, and most cases mirror the same type of mole as the preceding pregnancy (Garrett, 2008; Sebire, 2003). Furthermore, following two episodes of molar pregnancy, 23 percent of later conceptions result in another molar gestation (Berkowitz, 1998). For this reason, women with a prior history of GTD should undergo first-trimester sonographic examination in subsequent pregnancies. Familial molar pregnancies, however, are rare (Fallahian, 2003).

Of other risk factors, combination oral contraceptive (COC) pill use has been associated with an increased risk of GTD. Specifically, prior COC use approximately doubles the risk, and longer duration of use also correlates positively with risk (Palmer, 1999; Parazzini, 2002). Moreover, women who used COCs during the cycle in which they conceived have a higher risk in some but not all studies (Costa, 2006; Palmer, 1999). Many of these associations, however, are weak and could be explained by confounding factors other than causality (Parazzini, 2002).

Some epidemiologic characteristics differ markedly between complete and partial moles. For example, vitamin A deficiency and low dietary intake of carotene are associated only with an increased risk of complete moles (Berkowitz, 1985, 1995; Parazzini, 1988). Partial moles have been linked to higher educational levels, smoking, irregular menstrual cycles, and obstetric histories in which only male infants are among the prior live births (Berkowitz, 1995; Parazzini, 1986).

Hydatidiform moles are abnormal pregnancies characterized histologically by aberrant changes within the placenta. Classically, the chorionic villi in these placenta show varying degrees of trophoblast proliferation and edema of the stroma within villi (Fig. 37-1). Hydatidiform moles are categorized as either complete hydatidiform moles or partial hydatidiform moles (Table 37-2). Chromosomal abnormalities play an integral role in hydatidiform mole development.

Complete Hydatidiform Mole

These molar pregnancies differ from partial moles with regard to their karyotype, their histologic appearance, and their clinical presentation. First, complete moles typically have a diploid karyotype, and 85 to 90 percent of cases are 46,XX. The chromosomes, however, in these pregnancies are entirely of paternal origin, and thus, the diploid set is described as diandric. Specifically, complete moles are formed by androgenesis, in which the ovum is fertilized by a haploid sperm that then duplicates its own chromosomes after meiosis (Fig. 37-2) (Fan, 2002; Kajii, 1977). The ovum fails to contribute chromosomes. Most of these moles are 46,XX, but dispermic fertilization of a single ovum, that is, simultaneous fertilization by two sperm, can produce a 46,XY karyotype (Lawler, 1987). Although nuclear DNA is entirely paternal, mitochondrial DNA remains maternal in origin (Azuma, 1991).

Microscopically, complete moles display enlarged, edematous villi and abnormal trophoblastic proliferation. These changes diffusely involve the entire placenta (see Fig. 37-1). Macroscopically, these changes transform the chorionic villi into clusters of vesicles with variable dimensions. Indeed, the name hydatidiform mole literally stems from this “bunch of grapes” appearance. In these pregnancies, no fetal tissue or amnion is produced. As a result, this mass of placental tissue completely fills the endometrial cavity (Fig. 37-3).

Clinically, the presentation of a complete mole has changed considerably. In the 1960s and 1970s, more than half of affected patients had anemia and uterine sizes in excess of that predicted for their gestational age. In addition, hyperemesis gravidarum, preeclampsia, and theca-lutein cysts developed in approximately one quarter of women (Soto-Wright, 1995). As described in Chapter 9, theca-lutein cysts develop with prolonged exposure to luteinizing hormone (LH) or β-hCG (Fig. 37-4). These cysts range in size from 3 to 20 cm, and most regress with falling β-hCG titers after molar evacuation. If such cysts are present, and especially if they are bilateral, the risk of postmolar GTN is increased.

Complete moles, however, infrequently present today with these traditional signs and symptoms (Mangili, 2008). As a result of β-hCG testing and sonography, the mean gestational age at evacuation currently approximates 12 weeks, compared with 16 to 17 weeks in the 1960s and 1970s (Drake, 2006; Soto-Wright, 1995). A large proportion of patients are asymptomatic at diagnosis (Joneborg, 2014). For the remainder, vaginal bleeding remains the most common presenting symptom, and β-hCG levels are often greater than expected. One quarter of women will present with uterine size greater than dates, but the incidence of anemia is less than 10 percent. Moreover, hyperemesis gravidarum, preeclampsia, and symptomatic theca-lutein cysts are now rare (Soto-Wright, 1995). Currently, these sequelae typically develop chiefly in patients without early prenatal care who present with a more advanced gestational age and markedly elevated serum β-hCG levels. Last, plasma thyroxine levels are often increased in women with complete moles, but clinical hyperthyroidism is infrequent. In these circumstances, serum free thyroxine levels are elevated as a consequence of the thyrotropin-like effect of β-hCG (Chap. 15).

Partial Hydatidiform Mole

These moles differ from complete hydatidiform moles clinically, genetically, and histologically. The degree and extent of trophoblastic proliferation and villous edema are decreased compared with those of complete moles. Moreover, most partial moles contain fetal tissue and amnion, in addition to placental tissues.

As a result, patients with partial moles typically present with signs and symptoms of an incomplete or missed abortion. Many women will have vaginal bleeding. However, because trophoblastic proliferation is slight and only focal, uterine enlargement in excess of gestational age is uncommon. Similarly, preeclampsia, theca-lutein cysts, hyper­thyroidism, or other dramatic clinical features are rare. Preevacuation β-hCG levels are typically much lower than those for complete moles and often do not exceed 100,000 mIU/mL. For this reason, partial moles are often not identified until after a histologic review of a curettage specimen.

Partial moles have a triploid karyotype (69,XXX, 69,XXY, or less commonly 69,XYY) that is composed of one maternal and two paternal haploid sets of chromosomes (see Fig. 37-2) (Lawler, 1991). The coexisting fetus present with a partial mole is nonviable and typically has multiple malformations with abnormal growth (Jauniaux, 1999).

Diagnosis

Clinical Assessment

In reproductive-aged women with vaginal bleeding, diagnoses may include gynecologic causes of bleeding and complications of first-trimester pregnancy. The trophoblast of molar pregnancies produce β-hCG, and elevated hormone levels reflect their proliferation. Accordingly, initial urine or serum β-hCG measurement and transvaginal sonography are invaluable in guiding evaluation. Because of these, first-trimester diagnosis of hydatidiform mole is now common.

Although β-hCG levels are helpful, the diagnosis of molar pregnancy is more frequently found sonographically. Most first-trimester complete moles demonstrate a complex, echogenic, intrauterine mass containing many small cystic spaces, which reflect swollen chorionic villi. Fetal tissues and amnionic sac are absent (Fig. 37-5) (Benson, 2000). In contrast, sonographic features of a partial molar pregnancy include a thickened, hydropic placenta with a concomitant fetus (Zhou, 2005).

However, there are diagnostic limitations. For example, β-hCG levels in early molar pregnancies may not always be elevated in the first trimester (Lazarus, 1999). Moreover, sonography can lead to a false-negative diagnosis if performed at very early gestational ages, before the chorionic villi have attained their characteristic vesicular pattern. Studies show that only 20 to 30 percent of patients may have sonographic evidence to indicate a partial mole (Johns, 2005; Lindholm, 1999; Sebire, 2001). Consequently, diagnosis in early gestations is usually difficult. Often, the diagnosis commonly is not made until after histologic review of the abortal specimen. In unclear cases with a live fetus and a desired pregnancy, fetal karyotyping to identify a triploid fetal chromosomal pattern can clarify the diagnosis and management.

Histopathology

The histopathologic changes typical of hydatidiform moles are listed in Table 37-2. But, in early pregnancy, it may be difficult to distinguish among these and a hydropic abortus. Hydropic abortuses are pregnancies formed by the traditional union of one haploid egg and one haploid sperm but are pregnancies that have failed. Their placentas display hydropic degeneration, in which villi are edematous and swollen, and thus mimic some villous features of hydatidiform moles (Fig. 37-6). Although no single criterion distinguishes these three, complete moles generally have two prominent features: (1) trophoblastic proliferation and (2) hydropic villi. In gestations younger than 10 weeks, however, hydropic villi may not be apparent, and molar stroma may still be vascular (Paradinas, 1997). As a result, identification of early complete moles must rely on more subtle histologic abnormalities, supplemented by immunohistochemical and molecular diagnostic techniques. Partial moles are optimally diagnosed when three or four major diagnostic criteria are demonstrated: (1) two populations of villi, (2) enlarged, irregular, dysmorphic villi (with trophoblast inclusions), (3) enlarged, cavitated villi (≥3 to 4 mm), and (4) syncytiotrophoblast hyperplasia/atypia (Chew, 2000). Good diagnostic reproducibility can still be achieved in most circumstances using these histologic distinctions of complete and partial mole.

Ancillary Techniques

Histopathologic evaluation can be enhanced by immunohistochemical staining for p57 expression and by molecular genotyping. p57KIP2 is a nuclear protein whose gene is paternally imprinted and maternally expressed. This means that the gene product is produced only in tissues containing a maternal allele. Because complete moles contain only paternal genes, the p57KIP2 protein is absent in complete moles, and tissues do not pick up this stain (Merchant, 2005). In contrast, this nuclear protein is strongly expressed in normal placentas, in spontaneous pregnancy losses with hydropic degeneration, and in partial hydatidiform moles (Castrillon, 2001). Accordingly, immunostaining for p57KIP2 is an effective means to isolate complete mole from the diagnostic list. For distinction of a partial mole from a nonmolar hydropic abortus, both of which express p57, molecular genotyping can be used. Molecular genotyping determines the parental source of polymorphic alleles. Thereby, it can distinguish among a diploid diandric genome (complete mole), a triploid diandric-monogynic genome (partial mole), or biparental diploidy (nonmolar abortus) (Ronnett, 2011).

Treatment

Suction curettage is the preferred method of evacuation regardless of uterine size in patients who wish to remain fertile (American College of Obstetricians and Gynecologists, 2014; Tidy, 2000). Nulliparous women are not given prostanoids to ripen the cervix since these drugs can induce uterine contractions and might increase the risk of trophoblastic embolization to the pulmonary vasculature (Seckl, 2010). Hysterectomy is rarely recommended unless the patient wishes surgical sterilization or is approaching menopause (Elias, 2010). Symptomatic theca-lutein ovarian cysts are an unusual finding and tend to regress after molar evacuation. In extreme cases, these may be aspirated, but oophorectomy is not performed except when torsion leads to extensive ovarian infarction (Mungan, 1996).

Prior to surgery, patients are evaluated for associated medical complications. Fortunately, thyroid storm from untreated hyperthyroidism, respiratory insufficiency from trophoblastic emboli, and other severe coexisting conditions are rare. Because of the tremendous vascularity of these placentas, blood products should be available prior to the evacuation of larger moles, and adequate infusion lines established.

At the beginning of the evacuation, the cervix is dilated to admit a 10- to 12-mm plastic suction curette. As aspiration of molar tissues ensues, intravenous oxytocin is given. At our institution, 20 units of synthetic oxytocin (Pitocin) are mixed with 1 L of crystalloid and infused at rates to achieve uterine contraction. In some cases, intraoperative sonography may be indicated to help reduce the risk of uterine perforation and assist in confirming complete evacuation. Finally, a thorough, gentle curettage is performed.

Following curettage, because of the possibility of partial mole and its attendant fetal tissue, Rh immune globulin is given to nonsensitized Rh D-negative women. Rh immune globulin, however, may be withheld if the diagnosis of complete mole is certain (Fung Kee, 2003).

Postmolar Surveillance

Gestational trophoblastic neoplasia develops after evacuation in 15 percent of patients with complete moles (Golfier, 2007; Wolfberg, 2004). Despite the trend of diagnosing these abnormal pregnancies at earlier gestational ages, this incidence has not declined (Seckl, 2004). Of those women who develop GTN, three quarters have locally invasive molar disease and the remaining one quarter develop metastases. In contrast, GTN develops in only 4 to 6 percent of patients with partial moles following evacuation (Feltmate, 2006; Lavie, 2005). Malignant transformation into metastatic choriocarcinoma does occur after partial mole evacuation, but this is rare (0.1 percent) (Cheung, 2004; Seckl, 2000).

No pathologic or clinical features at presentation accurately predict which patients will ultimately develop GTN. Because of the trophoblastic proliferation that characterizes these neoplasms, serial serum β-hCG levels following molar evacuation can be used to effectively monitor patients for GTN development. Therefore, postmolar surveillance with serial quantitative serum β-hCG levels is the standard. Titers are monitored following uterine evacuation at least every 1 to 2 weeks until they become undetectable.

After undetectable β-hCG levels are achieved, subsequent monthly levels are drawn during 6 months of surveillance for all patients with a molar gestation (Sebire, 2007). However, poor compliance with prolonged monitoring has been reported—especially among indigent women and certain ethnic groups in the United States (Allen, 2003; Massad, 2000). A single blood sample demonstrating an undetectable level of β-hCG following molar evacuation is sufficient to exclude the possibility of progression to GTN in most patients. Thus, some women, especially those with a partial mole, may be safely discharged from routine surveillance once an undetectable value is achieved (Lavie, 2005; Wolfberg, 2004). Shortened surveillance could enable women to attempt a subsequent pregnancy sooner. However, GTN may still rarely develop after an hCG level has returned to normal, potentially leading to increased morbidity (Kerkmeijer, 2007; Sebire, 2007).

When pregnancies occur during the monitoring period, the resulting normal β-hCG production can hinder detection of postmolar progression to GTN (Allen, 2003). But other than complicating the monitoring schedule, these pregnancies fortunately are otherwise uneventful (Tuncer, 1999). To prevent difficulties with interpretation, women are encouraged to use effective contraception until achieving a β-hCG titer <5 mIU/mL or below the individual assay’s threshold. COC use decreases the likelihood of pregnancy compared with less effective barrier contraception and does not increase the risk of GTN (Costa, 2006; Gaffield, 2009). Injectable medroxyprogesterone acetate is particularly useful when poor compliance is anticipated (Massad, 2000). In contrast, intrauterine devices are not inserted until the β-hCG level is undetectable because of the risk of uterine perforation if an invasive mole is present.

Prophylactic Chemotherapy

At the time of molar evacuation, chemotherapy can be administered to help prevent GTN development in high-risk patients who are unlikely to be compliant or for whom β-hCG surveillance is not available. In clinical practice, however, correctly categorizing a woman as high-risk for GTN development is difficult, as no combination of risk factors is universally accepted. Typical patients have complete moles and multiple risk factors, such as age >40 years, previous history of molar pregnancy, or an excessively high β-hCG titer prior to evacuation. That said, few women are ultimately assigned to this group. Moreover, due to the risks of increased drug resistance, delayed treatment of GTN, and toxic side effects, this practice cannot currently be recommended (American College of Obstetricians and Gynecologists, 2014; Fu, 2012). Prophylactic chemotherapy is not routinely offered in the United States and Europe.

However, a single dose of dactinomycin has been shown to reduce the incidence of postmolar GTN in certain populations. For example, in one randomized trial, 60 Thai women who had high-risk complete moles were assigned to receive either prophylactic dactinomycin or placebo at the time of evacuation (Limpongsanurak, 2001). Adjuvant chemotherapy reduced the incidence of GTN from 50 percent to 14 percent, but toxicity was significant. As a result, prophylactic chemotherapy is generally used only in those countries with limited resources to reliably monitor patients after evacuation (Uberti, 2009).

Ectopic Molar Pregnancy

The true incidence of GTD developing outside the uterine cavity approximates 1.5 per 1 million births (Gillespie, 2004). More than 90 percent of suspected cases will reflect an overdiagnosis of florid extravillous trophoblastic proliferation in the fallopian tube (Burton, 2001; Sebire, 2005b). As with any ectopic pregnancy, initial management usually involves surgical removal of the conceptus and histopathologic evaluation.

Coexistent Fetus

At times, a twin pregnancy can consist of a hydatidiform mole and a coexisting fetus. The estimated incidence is 1 per 20,000 to 100,000 pregnancies (Fig. 37-7). Sebire and associates (2002b) described the outcome of 77 twin pregnancies, each composed of a complete mole and a healthy cotwin. Of this group, 24 women chose to have an elective termination, and 53 continued their pregnancies. Twenty-three gestations spontaneously aborted at less than 24 weeks, two were terminated due to severe preeclampsia, and 28 pregnancies lasted at least 24 weeks—resulting in 20 live births. The authors demonstrated that coexisting complete moles and healthy cotwin pregnancies have a high risk of spontaneous abortion, but approximately 40 percent result in live births. The risk of progression to GTN was 16 percent in first-trimester terminations and not significantly higher (21 percent) in women who continued their pregnancies. Because the risk of malignancy is unchanged with advancement of gestational age, pregnancy continuation may be allowed, provided that severe maternal complications are controlled and fetal growth is normal. Importantly, these cases should be distinguished early from a single partial molar pregnancy with its abnormal associated fetus. Fetal karyotyping to confirm a normal fetal chromosomal pattern is also recommended (Marcorelles, 2005; Matsui, 2000).

This term primarily encompasses pathologic entities that are characterized by aggressive invasion of the endometrium and myometrium by trophoblast cells. Histologic categories include common tumors such as the invasive mole and gestational choriocarcinoma, as well as the rare placental-site trophoblastic tumor and epithelioid trophoblastic tumor. Although these histologic types have been characterized, in most cases of GTN, no tissue is available for pathologic study. Instead, GTN is diagnosed based on elevated β-hCG levels and managed clinically.

Gestational trophoblastic neoplasia typically develops with or follows some form of pregnancy. Most cases follow a hydatidiform mole. Rarely, GTN develops after a live birth, miscarriage, or termination. Occasionally, the antecedent gestation cannot be confirmed with certainty. Many of the reported nonmolar cases may actually represent disease originating from an unrecognized early mole (Sebire, 2005a).

Histologic Classification

Invasive Mole

This common manifestation of GTN is characterized by whole chorionic villi that accompany excessive trophoblastic overgrowth and invasion (Fig. 37-8). These tissues penetrate deep into the myometrium, sometimes to involve the peritoneum, adjacent parametrium, or vaginal vault. Such moles are locally invasive but generally lack the pronounced tendency to develop widespread metastases typical of choriocarcinoma. Invasive moles originate almost exclusively from a complete or a partial hydatidiform mole (Sebire, 2005a).

Gestational Choriocarcinoma

This extremely malignant tumor contains sheets of anaplastic trophoblast and prominent hemorrhage, necrosis, and vascular invasion (see Fig. 37-8). However, formed villous structures are characteristically absent. Gestational choriocarcinoma initially invades the endometrium and myometrium but tends to develop early blood-borne systemic metastases (Fig. 37-9).

Most cases develop following evacuation of a molar pregnancy, but these tumors may also follow a nonmolar pregnancy. Specifically, gestational choriocarcinoma develops in approximately 1 in 30,000 nonmolar pregnancies. Two thirds of such cases follow term pregnancies, and one third develop after a spontaneous abortion or pregnancy termination. One review of 100 patients with nonmolar gestational choriocarcinoma reported that 62 presented after a live birth, 6 after a live birth preceded by a molar pregnancy, and 32 after a nonmolar abortion (Tidy, 1995). Vaginal bleeding was the most common symptom in all groups. For this reason, abnormal bleeding for more than 6 weeks following any pregnancy warrants evaluation with β-hCG testing to exclude a new pregnancy or GTN.

When choriocarcinoma is diagnosed after a live birth, the antecedent pregnancy usually proceeded normally to term. One case series collected between 1964 and 1996 showed that in 89 percent of cases, the preceding pregnancy had produced an uncomplicated live birth (Rodabaugh, 1998). Hydrops, while a notable complication in the remaining fetuses in this earlier series, was not observed in a more recent cohort compiled between 1996 and 2011 (Diver, 2013). Occasionally, unanticipated choriocarcinoma is detected in an otherwise normal-appearing placenta at delivery. More commonly, however, the diagnosis of choriocarcinoma is delayed for months due to subtle signs and symptoms. Most patients present with intermenstrual bleeding, and high β-hCG levels are detected (Lok, 2006). Less frequently, the diagnosis is made in an asymptomatic woman by an incidental positive pregnancy test (Diver, 2013). In part because of the typical delay to diagnosis, choriocarcinomas following term pregnancies are associated with high-risk features and a higher mortality rate than GTN following nonmolar abortions. Death rates range from 10 to 15 percent (Diver, 2013; Lok, 2006; Rodabaugh, 1998; Tidy, 1995).

In contrast to this gestational choriocarcinoma, primary “nongestational” choriocarcinoma is an ovarian germ cell tumor (Chap. 36). Although rare, ovarian choriocarcinoma has a histologic appearance identical to that of gestational choriocarcinoma. It is in part distinguished by the lack of a preceding pregnancy (Lee, 2009).

Placental-site Trophoblastic Tumor

This tumor consists predominantly of intermediate trophoblasts at the placental site. It is a rare GTN variant with unique disease behavior. Placental-site trophoblastic tumor (PSTT) can follow any type of pregnancy but develops most commonly following a term gestation (Papadopoulos, 2002). Typically, patients have irregular bleeding months or years after the antecedent pregnancy, and the diagnosis is not entertained until endometrial sampling has been performed (Feltmate, 2001). PSTT tends to infiltrate only within the uterus, disseminates late in its course, and produces low β-hCG levels (van Trommel, 2013). Of interest, an elevated proportion of free β-subunit can help to discriminate it from other GTN types if the endometrial biopsy is equivocal (Cole, 2008; Harvey, 2008). When this tumor does spread, the pattern mirrors that of gestational choriocarcinoma. Metastases often spread to the lungs, liver, or vagina (Baergen, 2006).

Hysterectomy is the primary treatment for nonmetastatic PSTT due to its relative insensitivity to chemotherapy. In particularly motivated patients, fertility-sparing procedures have mixed results (Feltmate, 2001; Machtinger, 2005; Papadopoulos, 2002; Taylor, 2013b).

Metastatic PSTT has a much poorer prognosis than its postmolar GTN counterparts. As a result, aggressive combination chemotherapy is indicated. EMA/EP regimens of etoposide, methotrexate, and dactinomycin (actinomycin D) that alternate with etoposide and cisplatin (Platinol) are considered the most effective (Newlands, 2000). Radiation, however, may also have a role. The overall 10-year survival is 70 percent, but patients with metastases, especially stage IV disease, have a much poorer prognosis (Hassadia, 2005; Hyman, 2013; Schmid, 2009).

Epithelioid Trophoblastic Tumor

This rare trophoblastic tumor is distinct from gestational choriocarcinoma and PSTT. The preceding pregnancy event may be remote, or in some cases, a prior gestation cannot be confirmed (Palmer, 2008). Epithelioid trophoblastic tumor develops from neoplastic transformation of chorionic-type intermediate trophoblast. Microscopically, this tumor resembles PSTT, but the cells are smaller and display less nuclear pleomorphism. Grossly, epithelioid trophoblastic tumor grows in a nodular fashion rather than the infiltrative pattern of PSTT (Shih, 1998). Hysterectomy is again the primary treatment due to presumed chemoresistance and since the diagnosis is usually confirmed in advance by endometrial biopsy. More than one third of patients will present with metastatic disease and demonstrable chemoresistance to multiagent therapy, which portends a poor prognosis (Davis, 2015; Palmer, 2008).

Diagnosis

Most GTN cases are clinically diagnosed, using β-hCG evidence to identify persistent trophoblastic tissue (Table 37-3). Tissue is infrequently available for pathologic diagnosis, unless a diagnosis of placental-site or nongestational tumor is being considered. As a result, most centers in the United States diagnose GTN on the basis of rising β-hCG values or a persistent plateau of β-hCG values for at least 3 weeks. Unfortunately, uniformity is lacking in the definition of a persistent plateau. Additionally, the diagnostic criteria are less stringent in the United States than in Europe. This is partly because of concern that some patients may be lost to follow-up if stricter criteria are used.

When serologic criteria for GTN are met, a new intrauterine pregnancy is excluded using β-hCG levels that are correlated with sonographic findings. This is done especially if there has been a long delay in monitoring of serial β-hCG levels or noncompliance with contraception or both.

Assessment

Patients with GTN undergo a thorough pretreatment assessment to determine disease extent. The initial evaluation may be limited to pelvic examination, chest radiograph, and pelvic sonography or abdominopelvic computed tomography (CT) scanning. Although approximately 40 percent of patients will have micrometastases not otherwise visible on chest radiography, chest CT is not needed because these small lesions do not affect outcome (Darby, 2009; Garner, 2004). However, pulmonary lesions identified on chest radiograph should prompt CT of the chest and magnetic resonance (MR) imaging of the brain. Fortunately, central nervous system involvement is rare in the absence of neurologic symptoms or signs (Price, 2010). Positron emission tomography (PET) may occasionally be useful to evaluate occult choriocarcinoma or relapse from previously treated GTN when conventional imaging is equivocal or fails to identify metastatic disease (Dhillon, 2006; Numnum, 2005).

Staging

Gestational trophoblastic neoplasia is anatomically staged based on a system adopted by the International Federation of Gynecology and Obstetrics (FIGO) (Table 37-4 and Fig. 37-10). Patients at low risk for therapeutic failure are distinguished from those at high risk by using the modified prognostic scoring system of the World Health Organization (WHO) (Table 37-5). About 95 percent of patients will have a WHO score of 0 to 6 and will be considered to have low-risk disease (Sita-Lumsden, 2012). The remainder will have a score of 7 or higher and be assigned to the high-risk GTN group. For the most accurate description of affected patients, the Roman numeral corresponding to FIGO stage is separated by a colon from the sum of the risk factor scores, for example, stage II:4 or stage IV:9. This description best reflects disease behavior (Ngan, 2004). Women with high-risk scores are more likely to have tumors that are resistant to single-agent chemotherapy. They are therefore treated initially with combination chemotherapy. Although patients with stage I disease infrequently have a high-risk score, those with stage IV disease invariably have a high-risk score. Women diagnosed with FIGO stage I, II, or III GTN have a survival rate approaching 100 percent (Lurain, 2010).

Nonmetastatic Disease

Invasive moles arising from complete molar gestations make up most nonmetastatic GTN cases. Approximately 12 percent of complete moles develop locally invasive disease after evacuation, compared with only 4 to 6 percent of partial moles. Epithelioid trophoblastic tumor and PSTT are other rare causes of nonmetastatic GTN. Locally invasive trophoblastic tumors may perforate the myometrium and lead to intraperitoneal bleeding (Mackenzie, 1993). Alternatively, vaginal hemorrhage can follow tumor erosion into uterine vessels, or necrotic tumor may involve the uterine wall and serve as a nidus for infection. Fortunately, the prognosis is typically excellent for all types of nonmetastatic disease despite these possible manifestations.

Metastatic Disease

Choriocarcinomas originating from complete molar gestations account for most cases of metastatic GTN. Three to 4 percent of complete moles develop metastatic choriocarcinoma after evacuation. This event is rare following any other type of molar or nonmolar gestation. Choriocarcinomas have a propensity for distant spread and should be suspected in any woman of reproductive age with metastatic disease from an unknown primary (Tidy, 1995). Moreover, because of this tendency, chemotherapy is indicated whenever choriocarcinoma is diagnosed histologically.

Although many patients are largely asymptomatic, metastatic GTN is highly vascular and prone to severe hemorrhage either spontaneously or during biopsy. Heavy menstrual bleeding is a common presenting symptom. The most common sites of spread are the lungs (80 percent), vagina (30 percent), pelvis (20 percent), liver (10 percent), and brain (10 percent) (Fig. 37-11). Patients with pulmonary metastases typically have asymptomatic lesions identified on routine chest radiograph and infrequently present with cough, dyspnea, hemoptysis, pleuritic chest pain, or signs of pulmonary hypertension (Seckl, 1991). In patients with early development of respiratory failure that requires intubation, the overall outcome is poor. Hepatic or cerebral involvement is encountered almost exclusively in patients who have had an antecedent nonmolar pregnancy and a protracted delay in tumor diagnosis (Newlands, 2002; Savage, 2015b). These women may present with associated hemorrhagic events. Virtually all patients with hepatic or cerebral metastases have concurrent pulmonary or vaginal involvement or both. Great caution is used in attempting excision of any metastatic disease site due to the risk of profuse hemorrhage. Thus, this practice is almost uniformly avoided except in extenuating circumstances of life-threatening brainstem herniation or chemotherapy-resistant disease.

Treatment

Surgery

Most patients diagnosed with postmolar GTN have persistent tumor confined to the endometrial cavity and are treated primarily with chemotherapeutic agents. Repeat dilatation and curettage is generally avoided to prevent morbidity and mortality caused by uterine perforation, hemorrhage, infection, uterine adhesions, and anesthetic complications (American College of Obstetricians and Gynecologists, 2014). Accordingly, second evacuations are not typically performed in the United States unless patients have persistent uterine bleeding and substantial amounts of retained molar tissue. Repeat uterine curettage is a more standard part of postmolar GTN management in Europe. This practice reduces both the number of patients needing any further treatment and the number of courses in those who do require chemotherapy (Pezeshki, 2004; van Trommel, 2005). A second evacuation followed by continued surveillance, however, is a less attractive option, even for poorly compliant patients, than single-agent chemotherapy (Allen, 2003; Massad, 2000).

Hysterectomy may play several roles in GTN treatment. First, it may be performed to primarily treat PSTT, epithelioid trophoblastic tumors, or other chemotherapy-resistant disease. Second, severe uncontrollable vaginal or intraabdominal bleeding may necessitate hysterectomy as an emergency procedure (Clark, 2010). Because of these more extreme indications, most women undergoing hysterectomy have elevated pretreatment risk scores, unusual pathology, and higher mortality rates (Pisal, 2002). Finally, adjuvant hysterectomy decreases the total dose of chemotherapy needed to achieve clinical remission in low-risk GTN. Patients with disease apparently confined to the uterus who do not desire future fertility should be counseled about this option (Suzuka, 2001). However, the risk of GTN persistence after hysterectomy remains approximately 3 to 5 percent, and these patients should be monitored postoperatively (American College of Obstetricians and Gynecologists, 2014).

Residual lung metastases may persist in 10 to 20 percent of patients achieving clinical remission of GTN after chemotherapy completion. These patients do not appear to have an increased risk of relapse compared with those having normal chest radiographs or CT scans. Thus, thoracotomy is not usually necessary unless remission cannot otherwise be achieved (Powles, 2006). In general, the optimal patient to be counseled for thoracotomy will have stage III GTN, a preoperative β-hCG level <1500 mIU/mL, and a solitary lung nodule resistant to chemotherapy (Cao, 2009; Fleming, 2008).

Chemotherapy for Low-Risk GTN

Methotrexate

Most patients with hydatidiform mole who develop GTN are at low risk of chemotherapy resistance (score 0-6) (Seckl, 2010). Single-agent methotrexate is the most common treatment, and complete response rates range from 67 to 81 percent for variations of the two most common intramuscular (IM) methotrexate regimens (Table 37-6). Although bundled as low-risk disease, the highest cure rates occur in patients with the lowest WHO scores (0-1), and rates decline proportionally as WHO scores rise (Sita-Lumsden, 2012). Thus, patients with a WHO score of 6 should at least be considered for upfront combination therapy (Taylor, 2013a). Overall, 19 to 33 percent of women develop methotrexate resistance and are switched to other agents, described subsequently.

With methotrexate, the Gynecologic Oncology Group (GOG) conducted a prospective cohort dose-escalation study (protocol #79) of weekly administration that established a maximum dose of 50 mg/m² with minimal toxicity (Homesley, 1988, 1990). This regimen is continued weekly until β-hCG levels are undetectable, and then two or three additional weekly doses are given (Lybol, 2012). Alternatively, Charing Cross Hospital and University of Sheffield investigators currently use an 8-day alternating regimen of 50 mg IM methotrexate on treatment days 1, 3, 5, and 7, and oral folinic acid, 7.5 to 15 mg taken orally on days 2, 4, 6, and 8. Treatment is repeated every 2 weeks (Taylor, 2013a).

As discussed more fully in Chapter 27, methotrexate is a folic acid antagonist that inhibits DNA synthesis. Mild stomatitis is the most common side effect, but other serosal symptoms, especially pleurisy, develop in up to one quarter of patients treated with low-dose methotrexate. Pericarditis, peritonitis, and pneumonitis are infrequent (Sharma, 1999). Toxicity develops more frequently with the more intense 8-day regimens compared with weekly administration. This is despite routine folinic acid “rescue,” which is provided to protect normal mucosal and serosal cells (Chap. 27) (Gleeson, 1993).

Dactinomycin

Dactinomycin is less frequently used for the primary treatment of low-risk disease due to toxicity concerns, but it has superior efficacy as a single agent (Alazzam, 2012a; Yarandi, 2008). In a prospective GOG trial (protocol #174) of low-risk GTN, patients were randomly assigned to biweekly “pulse” 1.25-mg dose dactinomycin or to weekly methotrexate, 30 mg/m². Among 215 eligible patients, a complete response was observed in 69 percent given dactinomycin and in 53 percent given methotrexate. However, advocates of methotrexate have speculated that the unexpectedly low efficacy of methotrexate observed in this study may be due to subtherapeutic dosing. Moreover, those randomized to dactinomycin were twice as likely to develop alopecia and were the only patients to develop grade 4 toxicity, defined in Chapter 27 (Osborne, 2008). As yet, no trials have directly compared pulse dactinomycin and the widely used 8-day methotrexate regimen. Since survival rates are so high, methotrexate is usually tried first because most clinicians consider it to be the least toxic.

Patients who do not respond to an initial single-agent chemotherapeutic regimen fail to have persistently dropping β-hCG levels. These women should have their score recalculated using the modified WHO prognostic scoring system. Most women will still be considered low-risk and may be switched to a single-agent second-line therapy. Methotrexate-resistant GTN often responds to dactinomycin (Chapman-Davis, 2012; Chen, 2004). The GOG demonstrated a 74-percent success rate in a Phase II trial (protocol #176) of pulse dactinomycin as salvage treatment in 38 patients with methotrexate-resistant GTN (Covens, 2006). Etoposide is less commonly used in this setting but is also effective (Mangili, 1996). Patients initially treated with pulse dactinomycin who develop resistant GTN may still be successfully treated with the 5-day course of dactinomycin (Kohorn, 2002). Alternatively, single-agent methotrexate or etoposide is often effective in these cases (Matsui, 2005).

Chemotherapy for High-Risk GTN

Approximately 5 percent of GTN patients present with high-risk disease and usually have numerous metastases months or years after the causative pregnancy. Such patients are likely to develop drug resistance to single-agent chemotherapy (Seckl, 2010). Etoposide, methotrexate, and dactinomycin (actinomycin D) alternating with cyclophosphamide and vincristine (Oncovin) (EMA/CO) chemotherapy is a well-tolerated and highly effective regimen for high-risk GTN. It is considered the preferred treatment for most high-risk disease. Bower and associates (1997) reported a 78-percent complete remission rate in 272 consecutive women. Similarly, other investigators have observed a 71- to 78-percent complete response rate with the EMA/CO regimen (Escobar, 2003; Lu, 2008). Response rates are comparable whether patients are treated primarily or after failure of single-agent methotrexate and/or dactinomycin.

Patients with high-risk disease have an overall survival rate of 86 to 92 percent, although approximately one quarter become refractory to or relapse from EMA/CO (Bower, 1997; Escobar, 2003; Lu, 2008; Lurain, 2010). Secondary treatment usually involves platinum-based chemotherapy combined with possible surgical excision of resistant disease (Alazzam, 2012b). Newlands and colleagues (2000) reported an 88-percent survival rate among 34 patients by replacing the cyclophosphamide and vincristine component with etoposide and cisplatin (EMA/EP). EMA/EP is an effective option in patients resistant to EMA/CO, but paclitaxel (Taxol) plus cisplatin alternating with paclitaxel plus etoposide (TP/TE) has comparable efficacy and appears less toxic (Patel, 2010; Wang, 2008). Bleomycin, etoposide, and cisplatin (BEP) is another potentially effective regimen (Lurain, 2005).

High-risk patients with a large disease burden are at risk for early death with standard EMA/CO due to tumor-lysis related hemorrhage and clinical deterioration. In these selected circumstances, “induction low-dose etoposide-cisplatin” appears to reduce the mortality risk 10-fold (Alifrangis, 2013).

Brain Metastases

Patients with cerebral metastases may present with seizures, headaches, or hemiparesis (Newlands, 2002). Occasionally, they are moribund on arrival after not recognizing the significance of their symptoms or following an extended delay in diagnosis. In such extenuating circumstances, emergency craniotomy may be indicated to stabilize the patient and is followed by critical care support throughout the active phase of treatment (Savage, 2015b). In experienced centers, virtually all GTN-related deaths occur in stage IV patients with WHO risk scores of 12 or more (Lurain, 2010).

Fortunately, the cure rate for those with brain metastases is high if neurologic deterioration does not occur within the first weeks after diagnosis. The sequence of aggressive multimodality therapy is controversial but may include chemotherapy, surgery, and radiation. Savage and coworkers (2015b) reported an 85-percent survival rate among 27 patients treated from 1991 to 2013 by EMA/CO or EMA/EP with an enhanced intravenous dose (1 g/m²) of methotrexate combined with intrathecal methotrexate until β-hCG levels were undetectable. Whole-brain radiation therapy can also be an efficacious adjunct to combination chemotherapy and surgery but can induce permanent intellectual impairment (Cagayan, 2006; Schechter, 1998).

Posttreatment

Surveillance

Monitoring of patients with low-risk GTN consists of weekly β-hCG measurements until the level is undetectable for 3 consecutive weeks. This is followed by monthly titers until the level is undetectable for 12 months. Patients with high-risk disease are followed for 24 months due to the greater risk of late relapse. Patients are encouraged to use effective contraception, as outlined earlier, during the entire surveillance period.

Treatment Sequelae

Despite the favorable prognosis, patients and their partners carry pregnancy concerns for a protracted time (Wenzel, 1992). Sexual dysfunction is an underreported complication (Cagayan, 2008). These and other potential sequelae highlight the importance of a multidisciplinary approach to management (Ferreira, 2009).

Although patients may expect a normal reproductive outcome after achieving remission from GTD, some evidence suggests that adverse maternal outcomes and spontaneous abortion occur more frequently among those who conceive within 6 months of chemotherapy completion (Braga, 2009). Women having a pregnancy affected by a histologically confirmed complete or partial mole may be counseled that the risk of a repeat mole in a subsequent pregnancy approximates 1 percent (Garrett, 2008). Most will be of the same type of mole as the preceding pregnancy (Sebire, 2003). Women who become pregnant within 12 months postchemotherapy for GTN can be reassured of a likely favorable outcome, although the safest option is still to delay pregnancy for the full year (Williams, 2014). Pregnancy after combination EMA/CO chemotherapy for GTN also has a high probability of success and favorable outcome (Lok, 2003). All major cytotoxic treatments except methotrexate increase the risk of early menopause (Savage, 2015a).

In some cases, secondary tumors can develop as a result of cancer treatment. Etoposide-based combination chemotherapy has been associated with an increased risk of leukemia, colon cancer, melanoma, and breast cancer up to 25 years after treatment for GTN. An overall 50-percent excess risk was observed (Rustin, 1996). Etoposide is therefore reserved to treat patients who are likely to be resistant to single-agent chemotherapy and, in particular, those with high-risk metastatic disease.

Quiescent Gestational Trophoblastic Disease

Patients with persistent mild elevations (usually ≤50 mIU/mL) of true β-hCG may have a dormant premalignant condition if no tumor is identified by physical examination or imaging studies (Khanlian, 2003). In this instance, phantom β-hCG, described next, should also be conclusively excluded as a possibility. The low β-hCG titers may persist for months or years before disappearing. Chemotherapy and surgery usually have no effect. Hormonal contraception may be helpful in lowering titers to an undetectable level, but patients are closely monitored since metastatic GTN may eventually develop (Khanlian, 2003; Kohorn, 2002; Palmieri, 2007).

Phantom β-hCG

Occasionally, persistent mild elevations of serum β-hCG are detected that lead physicians to erroneously treat patients with cytotoxic chemotherapy or hysterectomy or both, when in reality no true β-hCG molecule or trophoblastic disease is present (Cole, 1998; Rotmensch, 2000). This “phantom” β-hCG reading results from heterophilic antibodies in the serum that interfere with the β-hCG immunoassay and cause a false-positive result.

Several steps can clarify the diagnosis. First, a urine pregnancy test can be performed. With phantom β-hCG, the heterophilic antibodies are not filtered or renally excreted. Thus, these test-altering antibodies will be absent from the urine, and urine testing will show true negative results for β-hCG. Importantly, to conclusively exclude trophoblastic disease by this method, the index serum β-hCG level must be significantly higher than the detection threshold of the urine test. Second, performing serial dilutions of the serum sample leads to a proportional decrease in the β-hCG level if β-hCG is truly present. However, phantom β-hCG measurements will be unchanged by dilution. In addition, if phantom β-hCG is suspected, some specialized laboratories may be able to block the heterophilic antibodies. Last, heterophilic antibodies will cause interference with one assay, but they may bind poorly to another assay’s antibodies. Thus, switching β-hCG assay kits to one by a different manufacturer may accurately demonstrate the absence of true β-hCG (Cole, 1998; Olsen, 2001; Rotmensch, 2000).