29 GESTATIONAL TROPHOBLASTIC NEOPLASIA
The term gestational trophoblastic neoplasia (GTN) encompasses a range of entities including the more common complete (CHM) and partial (PHM) hydatidiform mole, invasive mole (IM), coexistent complete mole and live fetus, and the rarer malignant entities choriocarcinoma (CC) and placental site trophoblastic tumor (PSTT). The major clinical significance of the diagnosis of CHM or PHM is the associated increased risk of development of subsequent persistent gestational trophoblastic neoplasia (pGTN), in the form of any of the other subtypes1 (Fig. 29-1). However, in many cases in clinical practice, the strict distinction between these entities is often artificial. Following a suspected or proven diagnosis of a molar pregnancy, surveillance is usually instituted with serum and urine human chorionic gonoadotropin (hCG) measurements and patients in whom the hCG levels either rise or fail to return to normal are considered as having pGTN and may be treated with chemotherapy without a further specific histologic tissue diagnosis having being made. The management of complications of GTN is most often performed in specialist tertiary referral centers. The aim of this chapter is not to provide a thorough dissertation on all aspects of GTN, but rather to discuss an overview of the important issues and to specifically highlight the use of ultrasound examination in the diagnosis and management of these conditions. For purposes of simplicity, the chapter is broadly divided into two sections, focusing on hydatidiform mole (HM) in the first instance and then pGTD and gestational trophoblastic tumors.
FIGURE 29-1 Illustration demonstrating the various subtypes of gestational trophoblastic disease and persistent gestational trophoblastic disease. Complete hydatidiform mole in which numerous vesicles, representing hydropic chorionic villi, are seen filling the uterus. Partial hydatidiform mole in which a fetus is seen associated with patchy cystic, hydatidiform change of the placenta. Dichorionic twin pregnancy with a normal fetus and coexisting complete hydatidiform mole. In these cases, the sonogram demonstrates a normal-appearing placenta associated with the fetus, as well as tissue that has the characteristic appearance of a complete hydatidiform mole. Invasive mole in which penetration of molar villi, usually from a complete hydatidiform mole, occurs in the myometrium. Choriocarcinoma in which trophoblastic invasion, in the absence of chorionic villus structures, may extend into and beyond the myometrium, with the potential of metastases (in this case, the lung). PSTT in which the myometrium is infiltrated by trophoblastic cells, which may extend beyond the uterus.
(Illustration by James A. Cooper, MD, San Diego, CA.)
Hydatidiform moles essentially represent an abnormality of placental development as a consequence of overexpression of paternally derived genes and are therefore examples of abnormalities of imprinting. They are associated with structural placental abnormalities and fetal developmental defects, and are characterized pathologically from nonmolar pregnancies by the presence of abnormal trophoblast hyperplasia. They are simply divided into two major subtypes, CHM and PHM according to pathologic and genetic features.2
CHMs are almost always diploid, but in most cases, all genetic material is of paternal origin, either as a conse quence of genomic endoreduplication following monospermic fertilization of an anucleate oocyte or, more rarely, dispermic fertilization of an anucleate oocyte (Fig. 29-2). In rare cases, it has now been recognized that an unusual genetic variant of CHM, termed biparental CHM exists, in which the karyotype appears to be normal diploid with compliments from both parents, but the phenotype is identical to classic CHM, the defect appearing to be a result of abnormalities of imprinting with subsequent overexpression of the paternal genome.3
FIGURE 29-2 Patterns of fertilization to account for chromosomal origin of complete (46, XX) and triploid partial moles (XXY). In a complete mole, one or two sperm fertilize an egg that has lost its chromosomes. Partial moles are due to fertilization of an egg by one diploid, or two haploid sperm, depicted in this example as one 23,X and one 23,Y. Rarely, biparental complete moles are described, which are thought to be related to an imprinting defect.
(Modified with permission from Crum CP: The female genital tract. In Kumar V, Robbins SL, Cotran RS [eds]: Pathologic Basis of Disease, 7th ed. Philadelphia, WB Saunders, 2005. Copyright © 2005 Saunders, An Imprint of Elsevier.)
Partial hydatidiform moles are almost always triploid, with the additional set of genetic material also being paternally derived, almost always due to dispermic fertilization of an apparently normal oocyte. In PHM, the presence of a maternal genome is associated with less trophoblastic hyperplasia and a greater degree of fetal development. Thus, overexpression of paternally transcribed genes is likely to play a role in the development of PHM, whereas CMs reflect both overexpression of paternally transcribed and loss of maternally transcribed genes. All HM are associated with increased risk of development of pGTN, which is about 15% post CHM and 0.5% post PHM.4
Classically, these entities are clinically, as well as genetically, distinct, with PHM being associated with the presence of a fetus and coexistent patchy hydatidiform change of the placenta, whereas CHMs are not associated with fetal development and demonstrate diffuse placental change (see Fig. 29-1). However, with increasing understanding of these pathologies, in combination with changes in the management of early pregnancy complications, such simplistic distinctions are no longer correct, because it is now recognized that histologic evidence of early fetal-type development can sometimes be identified in cases of CHM,5 whereas the majority of cases of PHM are associated with early pregnancy failure in the absence of an identifiable fetus.6 Therefore, it is important to emphasize that the classic clinical, sonographic and histopathologic descriptions of these conditions are now outdated in the era of current gynecologic management, and revised and updated criteria for their diagnosis are required.
The gold standard for the definitive diagnosis, and subtyping, of HM is histopathologic examination by a specialist pathologist, because both CHM and PHM have distinctive histologic characteristics even in the first trimester.7–9 In the normal placenta, cytotrophoblast, syncytiotrophoblast, and intermediate trophoblast are present during placental development. Villus trophoblast covers chorionic villi and consists predominantly of cytotrophoblast and syncytiotrophoblast, cytotrophoblast being the proliferating component while syncytiotrophoblast differentiates from fusion of cytotrophoblast, composed of multinucleated cells with abundant dense cytoplasm and secretion of hCG. The diagnostic pathologic features of molar pregnancies are essentially characterized by abnormal proliferation of villus trophoblast. (Fig. 29-3). Intermediate trophoblast, of various subtypes, is also present, and although of prime importance in implantation and placental site trophoblastic tumors, is not a major component of the diagnostic criteria for molar pregnancies.
FIGURE 29-3 Photomicrographs of the various subtypes of gestational trophoblastic disease. A. Hydatidiform mole. Photomicrograph of a first trimester complete hydatidiform mole demonstrating characteristic budding architecture, with collapsed stromal blood vessels (thin arrow) and multifocal abnormal trophoblastic hyperplasia (thick arrow). B. Partial hydatidiform mole. Photomicrograph of a first trimester partial hydatidiform mole demonstrating irregular villus outlines, with trophoblastic pseudoinclusions (thin arrow) and focal abnormal trophoblastic proliferation (thick arrow). C. Invasive mole. Photomicrograph of an invasive mole demonstrating molar villi (arrow) infiltrating between myometrial (M) bundles. (H & E, original magnification × 40.) D. Choriocarcinoma. Photomicrograph of fragments of tissue from choriocarcinoma demonstrating a focal bilaminar pattern, recapitulating cytotrophoblast, and syncytiotrophoblast phenotypes (thin arrow), with pleomorphic syncytiotrophoblastic differentiation (thick arrow). E. Placental site trophoblastic tumor. Photomicrograph of a placental site trophoblastic tumor demonstrating infiltration by predominant mononuclear cells (thin arrow) with nondestructive surrounding of vessels (thick arrow).
In the classic, although now rarely seen, second trimester CHM, there is marked villus hydrops with extensive circumferential villus trophoblast hyperplasia and central villus cistern formation (Figs. 29-3 and 29-4). However, in the first trimester, such features are not yet developed, the diagnostic features being abnormally distributed villus trophoblast hyperplasia, relative lack of villus hydrops, sheets of pleomorphic extravillous trophoblast, collapsed villus blood vessels and marked stromal karyorrhectic debris in association with characteristic abnormal budding villus architecture. In addition, extravillous trophoblast invasion may be abnormal in CHM, with absence of normal controlled endovascular trophoblast plugging of decidual vessels, instead showing florid interstitial trophoblast invasion, sometimes resulting in interstitial hemorrhage.7–10
(Courtesy of Jonathan Carter, MD, The University of Sydney, Sydney, Australia.)
PHMs also exhibit nonpolar trophoblast hyperplasia, but in first trimester PHMs, this may be patchy and mild. Other features include patchy mild villus hydrops with focal cistern formation, numerous villus vessels containing nucleated fetal red cells, angiomatoid change, abnormally shaped scalloped or dentate villi, with trophoblastic pseudoinclusions and villus stromal fibrosis. Fetal tissue may also be present. The main diagnostic histopathologic features of CM, PM, and nonmolar hydropic abortion (HA) in products of conception evacuated in the first trimester are summarized in Table 29-1 (see Fig. 29-3).
It should be noted that although in most cases histopathologic diagnosis is easy, in some cases, a definite diagnosis may not be possible, particularly in cases with only limited material submitted for examination, and conceptions affected by other aneuploidies that may appear histologically similar to PHM. In such cases, ancillary diagnostic techniques may be required, including p57KIP2 immunohistochemistry11,12 (absent staining in CHM, positive nuclear staining in other diagnoses), assessment of ploidy using in situ hybridization or flow cytometry,13 and microsatellite polymorphism analysis.14
Classically, patients with singleton CHM present with vaginal bleeding, uterine enlargement greater than expected for gestational age, and abnormally high levels of serum hCG levels with passage of vesicles per vagina (see Figs. 29-3 and 29-4). Medical complications include pregnancy-induced hypertension, hyperthyroidism, hyperemesis, anemia, and development of ovarian theca-lutein cysts, which may even occasionally lead to ovarian torsion or cyst rupture15 (Fig. 29-5). This classic presentation now is highly unusual,16 because the majority of uncomplicated pregnancies, and all those presenting with first-trimester vaginal bleeding, now undergo routine ultrasound examination, at which time either molar change or missed abortion is diagnosed, both of which lead to evacuation of the uterus. This has resulted in a marked reduction in the number of molar pregnancies presenting with the symptoms and signs of overgrowth of trophoblast and excessive hCG secretion, and most cases are now evacuated in the late first trimester, the average gestation at diagnosis in the United Kingdom now being approximately 10 weeks.17 It should be noted at this point, that ultrasound examination leads to earlier uterine evacuation in most cases following diagnosis of a failed pregnancy rather than identification of molar change (see later). In one recent study of patients with CHM, 40% were asymptomatic, the condition detected by routine sonographic examination, while 60% presented with vaginal bleeding; only 2% reported hyperemesis, and none had any other systemic manifestations.18
HMs have been recognized by ultrasound examination for many years, but much of the literature refers to ongoing pregnancies with second trimester moles. The typical features reported include a uterine cavity filled with central heterogeneous mass with anechoic spaces of varying size and shape, a snowstorm-like appearance in the early days of ultrasound, without associated fetal development (see Figs. 29-1 and 29-6).19,20 In addition, theca lutein cysts secondary to the high hCG levels were reported, producing either soap-bubble or spoke-wheel appearance of the ovaries, which are enlarged (see Figs. 29-5 and 29-7). Doppler ultrasound examination in this scenario almost always demonstrates high velocities and low resistance to flow in the uterine arterial circulation and low resistance flow from trophoblastic tissue (Fig. 29-8). In cases of invasive mole, in addition to the central uterine lesion, myometrial invasion is present, with choriocarcinoma (CC) appearing as a mass enlarging the uterus, with a heterogeneity correspondling to areas of necrosis and hemorrhage21 (Fig. 29-9).
FIGURE 29-6 Two patients with a second trimester hydatidiform mole. A. A large moderately echogenic mass with numerous cystic spaces (arrows) is seen filling the central uterine cavity. The cystic spaces likely represent the markedly hydropic chorionic villi. B. A large moderately echogenic mass is seen filling the uterine cavity. In addition to the typical cystic hydatidiform molar tissue, large areas of hemorrhage are seen.
FIGURE 29-7 A. Sonogram from a patient with bilateral theca lutein cysts. The typical multilocular appearance is noted in the left ovary. B. Intraoperative photograph of the postgravid uterus and ovaries in a patient with coexistant mole and live pregnancy. Bilateral theca lutein cysts are identified. The characteristic multilocular appearance is seen.
(From Yee B, Tu B, Platt LD: Coexisting hydatidiform mole with a live fetus presenting as a placenta previa on ultrasound. Am J Obstet Gynecol 144:726, 1982.)
However, in an analogous manner to the changing clinical aspects, such sonographic presentations are also now only rarely encountered, with most cases being evacuated in the first trimester, when ultrasound demonstration of hydatidiform changes is much less readily apparent.
Several studies have now reported on the use of ultrasonography for detection of HM in early pregnancy22,23 (Fig. 29-10). Fine et al24 reported on the sonographic features of 22 PHMs compared with nonmolar abortions and suggested that cystic changes and increased placental echogenicity may be useful diagnostic features. Lazarus et al25 examined 21 CHMs in the first and second trimester (4–18 weeks) and reported that correct pre-evacuation diagnosis of HM could be made in about half; in no cases were theca lutein cysts present (Fig. 29-11). More recently, Benson et al26 described 24 CHMs examined sonographically in a specialist center and reported that molar pregnancy was suspected by ultrasound in almost 80%, suggesting, as with all sonographic techniques, that experienced operators may achieve a higher rate of diagnostic accuracy. Jauniaux et al27 reported that 10 of 11 pregnancies with sonographic features suggestive of HM at 10 to 14 weeks of gestation were pathologically proven HMs, indicating that specificity of sonographic diagnosis is high, but allowing no comment on senstivity. Furthermore, Lindholm et al28 reported that, on the basis of sonography and macroscopic examination of products of conception for molar change, about 80% of complete moles and 30% of partial moles could be detected, but it was noted that in several cases, macroscopic exami nation following uterine evacuation rather than the sonogram raised the first clue to the diagnosis (Fig. 29-12). In most cases of both PHM and CHM reported in which the diagnosis was not suspected before evacuation, the initial sonographic diagnosis was that of missed miscarriage/anembryonic pregnancy.29 Two larger, recent studies have reported that correct pre-evacuation identification of molar pregnancy by ultrasound in the first and early second trimester is achieved in around 40% to 60% of cases.17,30 The largest study by far to specifically address this issue examined the accuracy of routine ultrasound examination in 1,053 consecutive patients referred to a UK Regional trophoblastic disease unit for histologic review for possible or probable HM17; because all cases of GTN, suspected clinically, sonographically, or on the basis of histopathologic findings, are registered at the center, this allowed the highest possible ascertainment. The median gestational age in this study was only 10 weeks (range 5–27 weeks; Fig. 29-13). The final review diagnosis was HM in 859 (82%) cases, including 253 (29%) CHM and 606 (71%) PHM. Nonmolar hydropic miscarriage was diagnosed following histologic review in 194 (18%). Overall, around 40% of HM cases had a pre-evacuation ultrasound diagnosis suggesting molar pregnancy, including about 80% of CHM and 30% of PHM, the remainder of cases appearing sonographically simply as a missed miscarriage (Fig. 29-14). There was a trend toward increasing ultrasound detection rate with increasing gestational age; sonographic features of HM were reported in 35% of cases before 14 weeks of gestation compared with around 60% after this time (Fig. 29-15). It should also be noted that more than 10% of the cases identified as molar on ultrasound examination were in fact nonmolar hydropic miscarriage on histologic review. The sensitivity, specificity, positive and negative predictive values for routine pre-evacuation ultrasound examination for HM therefore are 44%, 74%, 88%, and 23%, respectively. These data compare favorably with a similar, smaller, study reporting on cases from 5 years previously in which the overall pre-evacuation ultrasound detection rate was approximately 30%.6 This improvement is presumably as a consequence of increasing specialization of obstetric sonographers, increasing experience in the recognition of such pregnancies, and improved ultrasound technologies. It is likely that some further improvements will occur, but there are pathophysiologic reasons limiting the proportion of HM, especially PHM, that will demonstrate sonographically detectable features in the first trimester.8,9
FIGURE 29-10 Sagittal sonogram from a patient with a first trimester hydatidiform mole. In this case, an echogenic mass (large arrows) is seen filling the uterine cavity. Except for one small cyst (small arrow), there are virtually none of the cystic spaces that are characteristically seen in more advanced molar pregnancies. Bl, bladder; M, myometrium.
FIGURE 29-11 Three patients presented with vaginal bleeding and postive pregnancy tests. A. This patient has a hyperechogenic mass (arrows) filling the uterine cavity. The appearance is similar to that of a first trimester hydatidiform mole (compare with Fig. 29-10). This was a nonviable pregnancy (missed abortion). B. Sonogram in a patient with early pregnancy and vaginal bleeding. This pregnancy was nonviable. A small cystic area, likely representing hydropic change (arrow), is seen. Am, amnion. C. In this patient, an irregular fluid collection (arrow) was seen in the uterine cavity. This has the appearance of a nonviable gestation. At pathologic evaluation, there was evidence of trophoblastic proliferation consistent with a first trimester hydatidiform mole. Bl, bladder.
FIGURE 29-13 Gestational age distribution at evacuation of products of conception referred with suspected diagnosis of complete or partial hydatidiform mole to a regional center in 2004. The median gestational age at evacuation is 10 weeks (range: 5–27 weeks).17
FIGURE 29-14 Pre-evacuation detection rates of complete (CM) and partial (PM) hydatidiform molar pregnancies following routine sonographic examination. Less than half of all hydatidiform moles (overall) in whom an ultrasound examination had been carried out were identified as molar pre-evacuation, the detection of complete moles being significantly better than for partial moles.
FIGURE 29-15 Pre-evacuation detection rates of hydatidiform molar pregnancies by gestational age. There is no significant difference in detection rates across the age range, but there is a clear trend toward improved detection rates at 14 weeks’ gestation and beyond.