23
Trophoblastic Neoplasms
tioma. Some of these terms are still in use, even though
the latter two have been replaced as greater knowledge of the biology has been accumulating.
In the United States, gestation-related choriocarci- noma is said to occur with a frequency of about 1 per 40,000 pregnancies (Hertig & Mansell, 1956). In their oft-depicted diagram (Fig. 23.1), Hertig and Mansell estimated that the lesion was preceded by a complete hydatidiform mole (CHM) in 50%, an abortion in 25%, a normal pregnancy in 22.5%, and an ectopic pregnancy in 2.5%. There is wide geographic variation in its inci- dence as it is in that of its precursors. In a study of cho- riocarcinoma from Japan, five of eight choriocarcinomas followed complete moles, one followed a term pregnancy (the only patient who died), and two followed abortions (Fukunaga & Ushigome, 1993a). All were diploid by flow cytometry.
The tumor consists of solid sheets of cytotrophoblast and syncytium. Choriocarcinoma has the great propensity of vascular invasion (Fig. 23.2) and, consequently, many choriocarcinomas are hemorrhagic and friable (Figs. 23.3 and 23.4). So much blood may be present in some chorio- carcinoma metastases that one may have to search long for the tumor cells (Fig. 23.5). Characteristically, broad sheets of cytotrophoblast form the central portion of the tumor, the periphery being syncytium. Nuclear pleomor- phism is common, but mitoses are confined to the cytotro- phoblast. Wolf and Michalopoulos (1992) studied the distribution of nuclear antigen in normal placentas and trophoblastic tumors. Its presence correlated with mitotic, reproductive activity and, in placentas, it was also con- fined to the cytotrophoblast, where it was strongly expressed. The same was true in CHM and choriocarcino- mas; the cytotrophoblast and extravillous trophoblast (X cells) stained, but not the syncytium. Several studies have shown that many intermediate trophoblastic cells are present in choriocarcinoma, even though they are not the hallmark of the lesion. These are the cells truly intermedi- ate between cytotrophoblast and syncytium and are not
837 Among trophoblastic neoplasms, choriocarcinoma is the
most well known. The classification of gestational tropho- blastic disease (GTD) has been alluded to in the previous chapter, and Soper et al. (1994), who evaluated 454 tumors according to three classifications, found the prognostic aspects to be fairly similar among these otherwise diver- gent classifications. Perhaps the clinical staging provided the best prognostication. An excellent and concise review of chorionic tumors has summarized the progress made in understanding these lesions and prescribing therapy for them (Berkowitz & Goldstein, 1996). A checklist for the characterization of trophoblastic malignancies has been provided by Lage (1999), and several other note- worthy reviews among many others are those by Silverberg and Kurman (1992), Baergen (1997), and Baergen and Rutgers (1997). The classification of GTD, recently endorsed by the World Health Organization (WHO) (Tavassoli et al., 2003), is the most widely used and is shown in Table 23.1.
Choriocarcinoma
Choriocarcinoma is a malignant neoplasm composed
exclusively of cytotrophoblast and syncytiotrophoblast. If
villi are associated with such invasive tumors, then the
lesion is referred to as “invasive mole” or chorioadenoma
destruens (Chapter 22). The nature of choriocarcinoma
was first correctly identified by Marchand (1895). Teach-
er’s detailed observations in 1903 followed Marchand’s
description. The early history of this tumor, the “decidu-
oma” as it was first known, has been admirably recounted
by Ober and Fass (1961). Until the neoplastic elements
of choriocarcinoma were ultimately shown to be derived
from the embryonic as opposed to maternal cells, their
derivation had been in dispute. Ewing (1910) was the first
investigator to differentiate between the various types of
trophoblastic neoplasms. He introduced or used terms
such as invasive mole, syncytial endometritis, and syncy-
the intermediate trophoblastic cells described in lesions of extravillous trophoblast (vide infra). They are espe- cially readily identified by electron microscopy (Pierce &
Midgley, 1963; Wynn & Davies, 1964) (Fig. 23.6). The X cells, the other major line of trophoblastic elements, the extravillous trophoblast, were identified with certainty in choriocarcinoma-like lesions; they are also not a striking feature of hydatidiform moles. Because of its dilated cyto- plasmic cisternae, the syncytium of choriocarcinomas is frequently vacuolated. This vacuolation may obscure the true nature of the cells. Much glycogen may also be con- tained in this tissue (Arkwright et al., 1993). When Cresci- manno et al. (1996) cloned BeWo choriocarcinoma cells (human choriocarcinoma cell lines) and identified two separate lines of cells, they found different behavioral ex- pression in these two lines. One was rapidly invasive and produced tumors and metastases in nude mice, and the other did not. The invasive line produced much greater quantities of gelatinase A (and less gelatinase B), and this activity was found to be stimulated by laminin and matri- gel. This is a first approach for understanding the com-
Table 23.1. World Health Organization classification of gest- ational trophoblastic diseaseHydatidiform mole Complete mole Partial mole Invasive mole Metastatic mole Trophoblastic neoplasms
Choriocarcinoma
Placental site trophoblastic tumor Epithelioid trophoblastic tumor
Nonneoplastic, nonmolar trophoblastic lesions Placental site nodule and plaque
Exaggerated placental site Source: Tavassoli et al., 2003.
Figure 23.2. Choriocarcinoma of the uterus. Solid sheets of neoplastic cytotrophoblast with much nuclear pleomorphism and syncytium are intermixed with blood. H&E ¥ 160.
plexity of invasion of the placenta into the decidua at implantation, as well as for understanding the invasive properties of some trophoblastic tumors, but not others.
It is usually difficult to assign a prognosis from the
histologic picture of a choriocarcinoma. Some choriocar-
Figure 23.1. Frequency relations of choriocarcinoma to its various precursors. (Source: Hertig & Mansell, 1956, with permission.)Figure 23.3. Liver metastases of choriocarcinoma following a third hydatidiform mole. There was no tumor in the uterus.
hemorrhagic tumor nodules have caused massive enlargement of liver. The ruler is placed underneath a large tumor growth within a hepatic vein.
Figure 23.5. Same patient as Figures 23.3 and 23.4.
Metastatic choriocarcinoma is in the liver following hydatidiform mole. Cords of disrupted liver cells are seen at right; a large tumor mass is at left and consists primarily of cytotrophoblast, with some syncytial surface. H&E ¥ 160.
Figure 23.6. Electron micrograph of two types of cytotropho- blast. At top are uniform oval nuclei and abundant ribosomes, but few other cytoplasmic constituents. Below are two large cells with irregular nuclei and many more mitochondria. Des- mosomes (D) are seen in both. (Source: Wynn & Davies, 1964, with permission.)
cinomas (even metastatic lung lesions) have regressed spontaneously (see Bardawil & Toy, 1959; Rauter, 1968).
Deligdisch and her coworkers (1978) suggested that solid tumor nests, with high pleomorphism and mitotic activity, have the worst prognosis. Fibrin deposits at the interface between tumor and host tissues are of good prognostic value. Another feature that bespeaks a better prognosis is a significant cellular “reaction” around the tumor (Elston, 1969). Sebire et al. (2004) have endeavored to use immunohistochemistry staining for p57
KIP2, a mater- nally expressed gene CDKN1C to differentiate chorio- carcinoma derived from moles or nonmolar pregnancies.
They found, however, that this marker does not provide a sufficiently accurate means.
To be inclusive it must also be mentioned that nonges- tational choriocarcinomas have been described on rare occasion. Thus, Liu et al. (2001) presented a gastric cho- riocarcinoma that was associated with an adenocarci- noma. It secreted large quantities of gonadotropin, which decreased after initially successful therapy. But the tumor recurred, with elevation of gonadotropins, and eventually led to the patient’s demise. Although she had three preg- nancies in the past, no connection to these could be estab- lished, and the authors considered “dedifferentiation” as a possible mechanism of formation. Interestingly, in their literature review of extrauterine primary choriocarcino- mas, the authors found that they are more frequent in the Japanese population.
Choriocarcinoma metastases occur most commonly in the lung and in the brain, but many other organs may be involved. The distribution of metastases was well recorded by Ober et al. (1971). That presentation also excelled in the description of the macroscopic lesions and in the
microscopic appearance of choriocarcinoma. Vaginal sec- ondaries and metastases to the cervix occur relatively frequently (Marquez-Monter & Velasco, 1967; Martin et al., 1983). Acosta-Sison (1958), who studied primary metastases in 32 patients, found the lung to be involved in 44% and the vagina in 31%. Eventually, the lungs were the site of metastases in 94% and the vagina in 44%;
other organs were much less often involved. A relatively uncommon site is the kidney, although isolated cases have been reported (Jarrett & Pratt-Thomas, 1984; Soper et al., 1988). Some patients have been successfully treated, despite large tumor burdens. A patient whose third hyda- tidiform mole (Fig. 23.3) proved fatal had extensive metastases in the liver, but she died from an exsanguinat- ing hemorrhage that originated in a duodenal lesion. This location is otherwise a relatively uncommon site of metastasis.
Pulmonary dissemination of molar tissue occurs often, but such vascular tumor deportation may lie dormant for many years. Figures 23.7 to 23.10 illustrate the lung of a patient who died from pulmonary hypertension 4 years after delivery of a hydatidiform mole. Her pulmonary arteries had thrombi and numerous masses of pure tro- phoblast. Neoplastic tissue was present in many vessels, but it did not completely traverse the vascular wall. There was an intense chronic inflammatory reaction at these sites, suggesting a barrier to invasion (Fig. 23.10). Similar patients have been described by Seckl et al. (1991) but they were successfully treated. The hypertensive effects of tumor embolization to the lung in general have been discussed in detail by Veinot and colleagues (1992), who believed this to be an important and often unrecognized problem.
Figure 23.7. Lung sections from a patient with pul- monary hypertension and infarction 4 years after hydatidiform mole. Throughout this lung the arteries were obstructed by choriocarcinoma and thrombi.
(Courtesy of Dr. A.T. Hertig. Patient discussed by Bardawil & Toy, 1959.)
Pulmonary artery is filled with solid choriocar- cinoma metastasis. There is an inflammatory reaction in the vascular wall at the point of invasion. H&E ¥ 40.
Figure 23.9. Same patient as in Figures 23.7 and 23.8. A small amount of viable tumor is present at left, the remainder of the artery is filled with thrombus.
H&E ¥ 40.
Figure 23.10. Same patient as Figures 23.7 to 23.9.
Choriocarcinoma fills the artery and is enmeshed in thrombus. At points of invasion there is much chronic inflammation, including many plasma cells. No
“villous stroma” is produced. H&E ¥ 100.
The cellular response to choriocarcinoma has been studied in some detail by Elston (1969). He found variable degrees of lymphocytic, histiocytic, and plasmacellular reactions in response to invading tumor cells and corre- lated this with outcome. Patients with tumors that had the most intense reaction fared best with chemotherapeutic treatment. Elston also saw in the maternal cellular response an attempt of the host to “reject” the neoplasm.
This pulmonary complication of latent vascular tumor growth is apparently not uncommon. Fahrner et al. (1959) described a patient who died with cor pulmonale and endarterial tumor growth 5 years after deli very of a mole.
Spiegel (1964) had a similar case and reviewed 10 others from the literature. Excellent preservation of neoplastic tissue resulted from the ability of the neoplasm to grow within blood. Other cases have come from Dyke and Fink (1967); and Bagshawe and Nobel (1966) discussed the various types of pulmonary metastasis in some detail.
They paid special attention to the radiologic and cardiac aspects. The latter authors and Evans and Hendrickse (1965) accumulated large series of this intraarterial pul- monary tumor spread. Many of the cases were associated with acute pulmonary hypertension and cor pulmonale.
Diagnostic curettage for the treatment of moles may be the mechanism of dissemination of malignant tropho- blast with villous tissue. Indeed, villi are occasionally present in such embolic lesions. Bagshawe (1964) pointed out that the presence of villi in trophoblastic lesions
“cannot be taken as a guarantee that it will remain benign indefinitely.” We strongly agree with this statement but recognize the difficulty experienced by the practicing pathologist and the dilemma of diagnostic semantics.
Often an absolute diagnosis of chorionic malignancy cannot be made from curettings alone, especially when villi are present. Optimal care requires follow-up with serial gonadotropin titer determinations. Carlson and his colleagues (1984) described a patient who developed fatal pulmonary lesions immediately after normal preg- nancy, despite chemotherapy. Bagshawe (1988) is incor- rect, however, to state that the pulmonary lesions occur only in London, and that they have not been recognized during life. They are rarely considered, to be sure, unless preceded by a recent CHM.
Choriocarcinoma most commonly follows CHM and invasive moles. Hertig and Mansell (1956) estimated that only 18.5% of CHM terminate with choriocarcinoma.
Most patients pursue a benign course once the uterus has been evacuated. As Figure 23.1 shows, choriocarcinoma may also follow normal gestations. The neoplasm has then occasionally been found within otherwise normal placentas and is then designated choriocarcinoma in situ.
Thus, Fukunaga et al. (1995, 1996) observed two cases in the placentas of spontaneous abortuses, clearly arising from stem villi. Other cases of choriocarcinoma in situ are described below. Other origins of choriocarcinoma include gonadal teratomas and, rarely, teratomas at other
Figure 23.11. Invasive mole (chorioadenoma destruens). Slices of uterus show the invasive, hemorrhagic mass of tissue. It is composed of mostly solid trophoblast with a few villous remnants.
sites, but they are then not GTDs. It is unusual for cho- riocarcinoma to arise without a known antecedent neo- plasm. Suzuki et al. (1993) were able to determine the origin of their choriocarcinomas with DNA analysis. Two cases followed complete hydatidiform moles and had androgenetic profiles, whereas another had parental DNA and followed a full-term pregnancy. They concluded from nine cases that polymerase chain reaction (PCR) study of the DNA of these tumors allows determination of the androgenetic origin. Seckl et al. (2000) made an important contribution when they studied the outcome of 3000 patients with partial hydatidiform mole (PHM).
Having verified the true triploidy in the PHMs studied, they found that 15 had required chemotherapy for per- sisting trophoblastic disease, and three of these women developed choriocarcinoma. It was the authors’ point to emphasize that PHMs also need serial follow-up in order to prevent this serious consequence.
Choriocarcinoma has no villi, in contrast to chorioad-
enoma destruens (invasive mole), in which villi arepresent. This arbitrary designation of the two lesions does
not signify any significant difference in their biologic
behavior, as chorioadenoma destruens may be signifi -
cantly invasive and destructive (Fig. 23.11). Indeed, the
two lesions are behaviorally essentially identical.
Choriocarcinomas have also been reported in some experimental animals, but their commonest antecedent, the CHM, has not been observed in nonhuman primate pregnancies. It is unknown if these animal choriocarcino- mas have the same unusual genetic background as some of the human gestational choriocarcinomas. There are several large reviews that discuss the major aspects of choriocarcinoma, including its historical perspectives, diagnosis, and treatment (Bagshawe, 1969; Ober et al., 1971; Park, 1971; Goldstein & Berkowitz, 1982; Silverberg
& Kurman, 1992).
Choriocarcinoma In Situ and Choriocarcinoma Coincident with Pregnancy
Driscoll (1963) described an incidental “choriocarcinoma”
in the mature, circumvallate placenta of a normal preg- nancy (Figs. 23.12 and 23.13). The mother and child did well. A small “infarct”-like lesion was sampled for histol- ogy and found to be a typical choriocarcinoma. The tumor invaded adjacent villi but not their vessels. Now that pla- centas are more often examined routinely, such tumors have been detected more commonly, and are always detected as infarcts on macroscopic study. Perhaps they should now be called placental choriocarcinoma. Brewer and Gerbie (1966) had two cases of choriocarcinoma developing within otherwise normal placentas. These
authors paid special attention to the earliest formation of this proliferative lesion and described the degenerative changes that occur within it. The degenerative process, they thought, may lead to detachment and deportation of tumor fragments. We have seen similar cases, three of which were the cause of massive transplacental fetal hemorrhage, and in one of them hydrops and fetal death (Santamaria et al., 1987) (Fig. 23.14). In another case, the anemic neonate had bled 250 mL into the maternal circu- lation, but survived. Mother and child had normal post- partum human chorionic gonadotropin (hCG) titers and remained well. The lesion appeared to be an infarct on gross examination and was sampled only because of the neonatal anemia. In yet another patient the mother developed central nervous system (CNS) and pulmonary metastases 1 month after delivery; the neonate had anemia and elevated hCG levels but remained well. In a case published by Lele et al. (1999), the 2 cm ¥ 1.5 cm incidental choriocarcinoma in a mature placenta led to disseminated metastases that were successfully treated.
It appeared as a typical infarct but was histologically a typical choriocarcinoma that had invaded adjacent normal villi. Metastatic choriocarcinomas identified during pregnancy have been seen more often. In a descrip- tion of a pertinent case, Barnes et al. (1982) found 18 cases described before 1970, and reviewed the literature at that time. Their patient, a gravida II (routine spontane- ous abortion 13 months earlier and with normal histol- ogy), suffered a metastatic choriocarcinoma in brain and lung in the second trimester. The fetus died and was devoid of metastases. The 270-g placenta contained a
Figure 23.12. Mature placenta with choriocarcinoma in situ (arrows).
This lesion was accidentally discov- ered and thought to be an infarct macroscopically. The patient did well. H&E ¥ 76. (Source: Driscoll, 1963, with permission.)
6 ¥ 6 ¥ 4 cm choriocarcinoma that had not metastasized to the stillborn fetus. Following chemotherapy and irra- diation, the woman had two normal pregnancies but suf- fered a glioblastoma from which she died 7½ years after the original tumor. Fox and Laurini (1988) found two additional cases of placental choriocarcinoma. In one, the fetus was stillborn, but the other did well. The placental lesion had the macroscopic appearance of an infarct. A particularly interesting case is that of Lage and Roberts (1993). Many white nodules in the placenta were diag- nosed as choriocarcinoma, and pulmonary metastases ensued that responded to therapy. Placental villi had not
been invaded. The 46,XX neonate remained normal.
Further, this publication brings together all of the litera- ture on choriocarcinoma in situ. It may be argued that these histologically identified lesions are not “malignant”
but, rather, in the sense used by Huber (1969), represent benign lesions, which he described as “chorionepitheliosis interna.” It may be so, except that not all such cases even- tuated without causing metastases. The concept of cho- riocarcinoma in situ is variably interpreted, and more must be learned from future biologic studies of early lesions. They are probably much more common than rec- ognized and fall, in a way, into the spectrum of confined
Figure 23.13. Same case as in Figure 23.12, but a higher power view of the tumorous elements.Note the obvious pleomorphism of tumor cells that surround the central villus. There is more syncytium than is usually present in choriocarci- noma. The dilated cisternae of the syncytium are obvious (bottom, center). H&E ¥ 200. (Source:
Driscoll, 1963, with permission.)
Figure 23.14. Choriocarcinoma in situ in an oth- erwise normal placenta. Solid nests of partially anaplastic tumor cells surround degenerating villi;
note also the invaded villi (arrow). Fetus was hydropic and stillborn due to transplacental he - morrhage. H&E ¥ 125. (Source: Santamaria et al., 1987, with permission.)
placental mosaicism. A relevant case is shown in Figure 23.15. This was a normal placenta, and random section disclosed this unusual villus with multiple layers of tro- phoblast and several mitoses. It is possible that such a lesion eventuates into a choriocarcinoma. The important case described by Heifetz and Czaja (1992) emphasized this. In their 69,XXX triploid PHM with fetus, a small nodule of truly malignant-appearing trophoblastic growth was found. Despite this, the woman pursued a normal postpartum course; the report also summarized the 10 or so choriocarcinomas in situ reported to 1992. Additional cases with fetomaternal hemorrhage or with a rise of a- fetoprotein (AFP) levels and benign subsequent course were presented by Duleba et al. (1992) and Ollendorff et al. (1990). Duleba et al., as did several others, treated their patient “expectantly” and showed in photographs that the lesion appeared grossly as an infarct. Ollendorff’s patient with elevated maternal serum AFP (MSAFP) and a placental choriocarcinoma had rising hCG titers and developed a pulmonary nodule. The discussion following this paper (Hustin & Jauniaux, 1991) proves that the lesion differed from their chorioangiocarcinoma (vide infra). Jacques et al. (1998) described five cases of intra- placental choriocarcinoma identified in a 5-year period.
In two women, pulmonary metastases were present, but none of the fetuses suffered disease, although two were premature and one had anemia. In situ hybridization for Y chromosomes was positive in the two gestations with male fetuses. The lesions had the gross appearances of infarcts. These authors also suggested that this intrapla- cental neoplasm may be more common than heretofore recognized.
Schopper and Pliess (1949) used the term chorionepi-
theliosis for borderline cases in a lengthy discussion oftrophoblastic properties. They were especially concerned with spontaneous cures of alleged choriocarcinomas and with ectopic chorionepitheliomatous growths. They described a woman with a vulvar trophoblastic tumor
that had developed during normal pregnancy. Benson et al. (1962) described massive fetomaternal hemorrhage in a patient who subsequently developed uterine choriocar- cinoma with pulmonary metastases. Blackburn (1976) and Feldman (1977) reported similar cases. In four addi- tional patients with small placental choriocarcinomas reported by Brewer and Mazur (1981), disseminated maternal lesions were found. The infants did not have tumors, but three were stillborn.
Hertig and Mansell (1956) indicated that choriocarci- noma follows (in the United States) normal pregnancies in a 1 per 160,000 incidence. It is thus not surprising that choriocarcinoma should also be found occasionally coin- cident with pregnancy and abortions (Fig. 23.1). Relevant cases are those by Heller and Householder (1952), Driscoll (1983), Miller et al. (1979), and Olive et al. (1984).
It is apparent from reviewing these reports that the choriocarcinoma that follows normal term pregnancy has a much poorer prognosis than that which follows CHM.
Patients with this complication of pregnancy also have often much earlier metastatic disease (Hutchison et al., 1968; Greene & McCue, 1978; Miller et al., 1979). In occa- sional cases of neonates with tumors, the placental cho- riocarcinoma was not detected or the placenta had not been examined. This was the case with cerebral chorio- carcinoma metastases that caused death in a 1-month-old child described by Chandra et al. (1990). The mother had no untoward sequelae and delivered another child, with normal placenta.
There are many other reports of choriocarcinoma that occur simultaneously in mother and infant or fetus.
Buckell and Owen (1954) described the death of a 7- week-old infant from chorionepitheliomatous metasta- ses; the mother needed a hysterectomy because she suffered postpartum choriocarcinoma. The authors referred to a similar case described by Emery in 1952.
Mercer et al. (1958) found a fatal choriocarcinoma in a 3-month-old infant whose mother died from dissemi-
Figure 23.15. A single villus within a term placenta, ran- domly sectioned, shows proliferation of trophoblast with several mitoses and hyperdiploidy, as judged by nuclear size.
Perhaps this is the precursor lesion of choriocarcinoma in situ. H&E ¥ 400.
nated disease. Another case, fatal for mother and infant, was reported by Daamen and colleagues (1961).
Witzleben and Bruninga (1968), who reported a case, sug- gested that these findings constituted a specific syndrome.
In their case, the mother had no obvious disease initially but later died from disseminated choriocarcinoma. Meta- static choriocarcinoma was the cause of fetal death in the report from Kruseman et al. (1977). The hydropic fetus had bled transplacentally. The mother developed meta- static disease and was successfully treated.
Chorangiocarcinoma
Yet another new tumor type was reported by Jauniaux and colleagues (1988). They called this lesion chorangio- carcinoma. It was described as a “missing link,” but they still believed that “it is a true malignant neoplasm or a voluminous chorangioma covered by extreme tropho- blastic hyperplasia” (Hustin & Jauniaux, 1991). The unique lesion they described appeared to be a solitary chorioangioma with extensive trophoblastic hyperplasia at its borders. The authors also concluded that chorangio- mas are not hamartomas but true neoplasms. Trask et al.
(1994) published a second case of chorangiocarcinoma, a choriocarcinoma in situ with accompanying chorangiosis, occurring in a twin pregnancy. The other twin’s placenta was normal and the mother did well. The lesion appeared to be an infarct macroscopically, as most other similar lesions have been. It introduces the question as to whether one should routinely sample all infarcts for histologic study, a practice that we have thus far avoided. The third case was described by Guschmann et al. (2003), again an incidental finding without deleterious consequences.
Even the angiogenetic factors that they studied in this case were normal; they postulated this to be a “collision tumor,” at least as being one possibility for the genesis of the tumor.
Exaggerated Placental Site:
Placental Site Nodule, Placental Site Trophoblastic Tumor, and Epithelioid Trophoblastic Tumor
Kurman et al. (1976) introduced the term trophoblastic
pseudotumor to designate a trophoblastic lesion they sawin 12 patients that behaved in a benign manner. The lesion tended to occur in young women, usually after pregnancy or abortion. The authors considered this entity to be “an exaggerated form” of the older entity syncytial endome- tritis. These placental site lesions were usually treated successfully by curettage. The trophoblastic pseudotumor, however, also formed occasional hemorrhagic nodules in the myometrium. This entity, we believe, is equivalent to the concept of chorionepitheliosis, which was discussed at
length by Schopper and Pliess (1949). It also probably encompasses the chorioma of Ewing (1910) and perhaps the syncytioma and other vague designations of border- line trophoblastic lesions discussed in the past. These enti- ties are not uncommon but were infrequently reported.
As better delineation became available, more cases were described, and a good survey is provided by Baergen and Rutgers (1997). The designation of these lesions depends much on the interpretation by individual pathologists and on the clinical behavior (see Fig. 23.19).
Because of the superficial resemblance to choriocarci- noma, it thus came as no surprise that metastatic conse- quences were later observed, despite the earlier belief that the trophoblastic pseudotumor was a benign growth.
Because several patients with apparently malignant lesions have since been identified, this led Scully and Young (1981) to introduce yet another term, the placental
site trophoblastic tumor (PSTT), which is characterizedby mononuclear extravillous trophoblast that infiltrates within the uterus and its vessels. Only rarely, are villi present. The absence of cytotrophoblastic cell masses and the usual absence of syncytiotrophoblast differentiates the entity from choriocarcinoma. The lesion is largely composed of placental site cells (extravillous trophoblast, or X cells) and blends with placental site nodules and plaques as well as exaggerated placental site (vide infra).
Such a lesion is shown in Figures 23.16 to 23.18. Here it is a nodule of trophoblast that had perforated the uterus 18 months after a normal term pregnancy. After hyster- ectomy, the patient thrived. There were no metastases.
Finkler and his colleagues (1988) treated seven patients with PSTTs and included a detailed review. One case fol- lowed a CHM, and the others occurred after abortion or premature delivery, in contrast to choriocarcinoma, which more commonly follows a CHM. All but one that fol- lowed an abortion pursued a benign course. Despite che- motherapy, this exceptional patient had metastatic disease in pelvic nodes. Although the authors did not publish any photomicrographs, they did describe an unprecedented high mitotic count. This point is particularly noteworthy because placental site cells (X-cells or extravillous trophoblast) are not usually very mitotic.
In the relatively short time since the description of
PSTT, a large number of cases have been reported, in
spite of the fact that it is the rarest proliferative lesion of
trophoblast. An interesting study of this lesion was under-
taken by Eckstein et al. (1985), who also reviewed the
literature most competently. In their view, differentiation
from choriocarcinoma should be easy because the lesion
is a “well-circumscribed yellowish mass in which hemor-
rhage and necrosis are less conspicuous than in chorio-
carcinoma.” The histology is more like a placental site
than like choriocarcinoma. The tumor in the aforemen-
tioned patient followed an apparently normal pregnancy
but was eventually fatal with widespread metastases. It
Figure 23.16. Placental site trophoblastic tumor that had penetrated the uterus 18 months after a normal, term delivery. Hysterectomy alone cured this patient. The “malignant” cells are pleomor- phic placental site giant cells (largely X cells).
There are no solid sheets of cytotrophoblast and few syncytial cells. There are no mitoses. H&E
¥ 100.
Figure 23.17. Placental site trophoblastic tumor;
same case as in Figure 23.16. H&E ¥ 160.
Figure 23.18. Placental site trophoblastic tumor.
H&E ¥ 250.
had a diploid DNA content, and biopsies were easily transplanted into nude mice. Unlike other lesions of this type, the tumor cells did not have a high mitotic rate.
Fukunaga and Ushigome (1993a,b) reviewed the entire topic comprehensively and added three cases of PSTT to the literature. One of their patients died with widespread metastases. That patient had a term delivery and returned 9 months later with uterine, vaginal, bladder, and lung tumors that resisted chemotherapy. Of the three cases they reported, two recovered, all had diploid flow cyto- metric DNA values, and all were of pregnancies with female babies. Characteristically, many fewer mitoses were seen in these lesions than in typical choriocarcino- mas, and histochemically they showed significantly less hCG production, and more human placental lactogen (hPL) staining despite their epithelioid appearance (Rhoton-Vlasak et al., 1998). As in eight other cases described, their fatal case had associated nephrotic syn- drome, a complication not reported in choriocarcinoma.
They estimated that perhaps 10% of PSTTs behave in a malignant manner.
Although histologic and clinical features of the tumor, such as tumor stage, advanced age, interval from previous pregnancy of greater than 2 years, previous term preg- nancy, clear cytoplasm, tumor necrosis, mitotic rate, and myometrial invasion, have been shown to be of prog- nostic value, no specific feature or features can predict benign versus malignant behavior (Rutgers et al., 1995, Baergen & Rutgers 1997). Furthermore, high mitotic rates have been identified in patients who ultimately pursued a benign course and low mitotic rates have been seen in patients with metastatic or even fatal disease (Fig. 23.19).
Rosenheim et al. (1980), who described a patient with uterine hemorrhage after abortion, decided that this
diagnosis can be made only on a hysterectomy specimen.
The uterus they studied showed an endometrial polyp that was composed of clot, with masses of atypical tro- phoblast interspersed. Villi had earlier been obtained by curettage from this patient. The tissue was like that of a CHM. The trophoblast invaded 30% of the myome- trium. After its removal, the patient experienced full recovery. Nagelberg and Rosen (1985) described other interesting biologic features in a patient following a normal pregnancy. They performed many endocrine studies, because the tumor was accompanied by viriliza- tion due to testosterone production, possibly secondary to ovarian theca cell stimulation. It was also relatively resistant to chemotherapy, a characteristic that distin- guishes PSTT from choriocarcinoma. After an eventual hysterectomy, the pathologist found a solitary 2-cm mass of tumor cells. This report provided much insight into other endocrine parameters of this condition. The pro- longed production of hPL by the X cells of such a lesion was held to be responsible for the erythrocytosis in a patient reported by Brewer et al. (1992). After hyster- ectomy the symptoms disappeared; the lesion was a 6-cm hemorrhagic mass that infiltrated the myometrium but was primarily composed of X cells. The difficulty of dif- ferential diagnosis was also highlighted by Horn et al.
(1997). They described a patient in whom a uterine lesion was initially mistaken for a cervical carcinoma.
Radical hysterectomy proved to have been unnecessary;
indeed, the lesion may merely have been an exaggerated placental site rather than a PSTT. Caution and experi- ence are necessary in the diagnosis before critical therapy is instituted. Another large cervical GTD (pre- sumably a choriocarcinoma) was treated only with che- motherapy and disappeared (Sohn et al., 1996). The only
Figure 23.19. Placental site trophoblastic tumor with epithelioid appearance and numerous mitoses.
H&E ¥ 400.
guides were hormone levels and the postgestational history of the lesion.
Fukunaga and Ushigome (1993a) believed that inter- mediate forms of these tumors exist. The case described by Hopkins et al. (1985) would thus perhaps still best be considered a choriocarcinoma following normal preg- nancy, rather than a placental site tumor. Its histologic composition most conformed to cytotrophoblast and syn- cytium, rather than to extravillous trophoblast. More precise methods to differentiate among these various lesions are desirable. Horn et al. (1996) presented differ- ential diagnostic criteria for these lesions. Such efforts may be accomplished by the use of histochemistry, as performed by Wells and Bulmer (1988) and Fukunaga and Ushigome (1993a,b). These authors found in one such lesion the existence of a few tumor cells with hCG markers; more cells contained hPL antigen (see also Rhoton-Vlasak et al., 1998). Duncan and Mazur (1989) compared the ultrastructure of one such tumor (follow- ing normal pregnancy) with that of nine choriocarcino- mas. The findings included an absence of cytotrophoblastic nests in the placental site tumor. The cellularity was com- posed primarily of intermediate trophoblast (extravillous trophoblast, X cells), with occasional interspersed syncy- tium. The authors found by immunohistochemistry that some of the neoplastic X cells contained hPL as well as hCG, which is not usually the case in normal cells. The cells in PSTT also stain positively for a-inhibin, Mel- CAM CD 146, a cell surface adhesion molecule and pla- cental alkaline phosphatase (Shih & Kurman, 1998b).
The most recently described neoplastic lesion of tro- phoblast is the epithelioid trophoblastic tumor. Shih and Kurman (1998b) described 14 cases of this unusual lesion that is characterized by sheets of large mononuclear cells that resemble epithelial carcinoma. The infiltrative neo- plasm is thought to derive from a specialized lineage of extravillous trophoblast and has a behavior more similar to that of migratory extravillous trophoblast, such as exist in the chorionic plate and extraplacental membranes.
Several patients have had metastases and some died from wide dissemination, but this tumor appears to have a similar behavior and prognosis as the usual type of pla- cental site trophoblastic tumor.
Syncytial endometritis is an ancient term that reflects
the presence of many trophoblastic giant cells in the floor of the placenta. Others might have used the term chorion-
epitheliosis to describe this morphology. This lesion isconsidered to be nonneoplastic and merely an exaggera- tion of a normal physiologic process. It is common in molar implantations. It has no relation to any inflammatory or infectious lesion, and the term has been abandoned in favor of exaggerated placental site (Tavassoli et al., 2003).
In reality, though, it is often impossible to distinguish this lesion from normal placental sites. One must also remem- ber that few normal placental sites ever come to be
observed by pathologists, as they are rarely curetted and death is uncommon at that stage of development (Dallen- bach-Hellweg, 1981). Therefore, the pathologist may be unduly alarmed by the appearance of a normal postpar- tum placental site. A relevant example is shown in Figures 23.20 and 21.21. It came from the curettage of a placental bed in an 8-week gestation and had no adverse conse- quences. But the exuberance of normal trophoblast at this site is apparent. Figure 23.22 illustrates another exagger- ated placental site. A cytogenetic NISH (nonradioactive in situ hybridization) study done by Faul et al. (1994) in cases of exaggerated placental site trophoblast indicated that the often-bizarre nuclei found in these lesions are poly- ploid or aneuploid, perhaps arising by cell fusion or by endoreduplication of chromosomes. Two cases of chorion- epitheliosis were published by Van Bogaert and Staquet (1977), presumably representing the same entity. The authors also considered the lesion to be a benign prolifera- tion of trophoblastic elements. To clarify the differential diagnosis, these authors developed a scheme for differen- tial diagnosis which has been adapted in Table 23.2.
Yet another lesion of extravillous trophoblast that has been described is the placental site nodule, which is a benign, nonneoplastic lesion that may be diagnosed many years after the preceding pregnancy. Nickels et al. (1978) described such a lesion in a 52-year-old woman whose last pregnancy was 20 years earlier. According to them, the lesion was “a normal anatomic event at the placental site after miscarriage or full-term pregnancy.” A review of 20 cases of placental site nodules and plaques was under- taken by Young and colleagues (1990), and 40 cases were
Figure 23.20. Exaggerated placental site at 8 weeks. The infil- trating cells split myometrial bundles and are composed of uni- or multinuclear extravillous trophoblast. This is a benign, normal placental floor but histologically may easily be confused with a placental site trophoblastic tumor. H&E ¥ 160.
Figure 23.21. Exaggerated placental site following fetal demise at 8 weeks. Endometrial glands are surrounded by extravillous trophoblast and these are admixed with multinucleated giant cells. H&E
¥ 250.
Figure 23.22. Curettings after normal pregnancy.
There was postpartum bleeding of this 35-year- old woman who had had eight pregnancies, with two live births. The lesion was interpreted as cho- riocarcinoma, but the patient had an uneventful course without chemotherapy. It is best inter- preted as exaggerated placental site. The cells are placental site giant cells (largely X cells). No solid sheets of cytotrophoblast are present, and there are no mitoses. H&E ¥ 260.
reviewed by Huetter and Gersell (1993). Most of the cells stained with antibodies to hPL and cytokeratin; all lesions were either incidental findings or followed with a benign course. There is now agreement that these nodules rep- resent retained placental sites in which much hyaliniza- tion and degenerative change are found. A useful marker for the identification of the proliferating cells is the pres- ence of major basic protein (MBP), and less useful is placental lactogen. Rhoton-Vlasak et al. (1998) found 100% of exaggerated placental site nodules to be positive for MBP, and 78% of PSTTs. Tsang et al. (1993) identified Mallory bodies in two cases, representing high molecular
weight cytokeratins. Shih et al. (1999) were able to dif- ferentiate the cells composing these lesions of extravil- lous trophoblast into distinct types—those presumably arising from the membranous extravillous trophoblast (mild proliferation, smaller, glycogen-rich) and those from the intervillous/placental floor spaces (more eosino- philic, larger, and pleomorphic). We found a nodule like this 5 years after the last pregnancy when a patient had hysterectomy for cervical carcinoma.
It is difficult to define these lesions precisely, especially
when the diagnosis depends on tissue obtained at curet-
tage (see also Horn et al., 1996; Horn & Bilek, 1997). A
detailed review with illustrations and prognostic predic- tions has been provided by Baergen and Rutgers (1997), but it remains a difficult topic for most pathologists. The hope of differentiating a potential malignancy among these placental site lesions by an assessment of ploidy or DNA has been disappointing. Most lesions have been diploid or tetraploid, as summarized by Lage and Sheikh (1997). To ascertain whether one could anticipate the development of persistent trophoblastic lesions, de Kaa et al. (1996) performed DNA analyses and cytogenetics.
They found this of limited value and also concluded that nuclear pleomorphism of CHM was of no prognostic significance.
Detailed study of genes by comparative genomic hybridization of four archival cases of PSTT undertaken by Hui et al. (2004) has shown chromosomal gain in two cases, no chromosomal losses, and neither gain nor loss in the remaining two cases. Initial results of studies by Shih and Kurman (1998a), who have been most inter- ested in defining these lesions, showed that double immu- nohistology using Ki-67 and Mel-CAM antibodies allows some differentiation of exaggerated placental site, PSTT, and choriocarcinoma.
Interesting attempts at the differential diagnosis have been published in the intervening years since the last edition of this book. First, in an exhaustive study, Oldt et al. (2002) showed with numerous markers that the cells that compose these gestational trophoblastic neoplasias (GTNs) (other than the choriocarcinoma) have fetal genetic and chromosomal markers; they are thus not maternal neoplasms. Further, all of these 42 tumors studied had heterozygous SNIPS (single nuclear poly- morphisms) and were thus different from the androge- netic moles and choriocarcinomas. It has been possible for Shih et al. (2004) to assign some of these lesions to different categories from a study of certain isoforms of the expression patterns of the p63 gene, and thus to assign putative ancestors among the trophoblast lineages, the
membranous vs. placental extravillous cytotrophoblast.
This has led to improved understanding of the otherwise confusing histopathology. In essence, then, this contribu- tion has confirmed their earlier deduction of the specific origin of these lesions (Shih & Kurman, 2001).
Ultrastructure of Trophoblastic Tumors
Wynn and Davies (1964) studied the fine structure of trophoblastic neoplasms that were transplanted to hamster cheek pouches. They combined their morpho- logic study with an endocrine assessment; they were unable to demonstrate estrogenic activity. Except for excessive dilatation of syncytial cytoplasmic channels, the tumor resembled early human trophoblast. Larsen et al. (1967) made similar observations and showed that syncytial cells performed phagocytosis. Other electron microscopy (EM) studies were carried out directly on human choriocarcinomas that had not been transplanted (Inferrera et al., 1967; Knoth et al., 1969a; Arai et al., 1976;
Duncan & Mazur, 1989). The various investigators made the same observations. Several of the descriptions included cells that are truly intermediate between cytotrophoblast and syncytium (“transitional” cells) (see Fig. 23.16), which was also shown in the choriocarcino- matous cell lines maintained in vitro by Knoth and col- leagues (1969b).
ANTIGENIC STUDIES OF TROPHOBLASTIC TUMORS
Most choriocarcinomas have at least some paternal genome. Perhaps most (at least those derived from hydatidiform moles) have only paternal chromosomes. It is therefore possible that the tumor’s antigenic deter- minants may engender immunologic rejection. This assumption was an important explanation for their frequently favorable prognosis. Elston (1969), though, described variably intense cellular reactions to the invading trophoblastic neoplasm and related it to outcome. Observa- tions such as this one have led to the study of immune interactions of the tumors with the host and to the determination of human leukocyte antigens (HLAs) in choriocarcinoma.
Mogensen and Kissmeyer-Nielsen (1968) studied HLA types in mothers, fathers, and their children in conjunction with placenta- associated choriocarcinoma. Despite antigenic differences between father and mother, most of the children were histocompatible with the mother. This compatibility, the authors suggested, made the tumors capable of growing in the uterus, rather than being rejected. Tomoda and his colleagues (1976) investigated HLAs and ABO antigens in patients with gestational trophoblastic neoplasia and their husbands.
ABO types had no correlation, but patients with choriocarcinoma were
“frequently incompatible” at various HLA types. Nevertheless, patients with choriocarcinoma were more frequently histocompatible with their husbands than those having invasive moles. Rudolph and Thomas (1971) had found significantly different HLAs in the mother and child of a patient who developed choriocarcinoma. The authors thereby refuted the notion that the tumors develop because of a failure of tro- phoblast to be rejected. Lawler et al. (1971) found that there was no increased histocompatibility between mothers and children in chorio- carcinoma cases. There was a suggestion that major incompatibilities of HLA and ABO types protected the mother from disseminated disease.
This aspect was further supported by findings of Mogensen and Kiss- meyer-Nielsen (1971). Ivaskova and her colleagues (1969) found lym- Table 23.2. Diagnostic differences between trophoblastic
pseudotumor and other diseases Exaggerated
Signs PSTT placental site Choriocarcinoma Mole
Blood vessel + - + -
invasion
Placental villi - + - +
Necrosis +/- +/- + -
Trophoblastic + + + +
invasion
Syncytial +/- + + +
trophoblast
Cytotrophoblast - + + +
Hemorrhage +/- - + -
pstt, placental site trophoblastic tumor.
phocytotoxic antibodies against HLAs in 12 of 13 patients with choriocarcinoma or moles and speculated that these antigens may be important in tumor rejection. Differences in the population of the HLA2
membrane attack complex of complement (MAC) did not exist from controls in a French population of choriocarcinoma patients (Amiel &
Lebovici, 1970). Ho et al. (1989) found that whites did not share HLAs, whereas Chinese with choriocarcinomas did. Yamashita et al. (1984) performed direct tissue studies to ascertain the possible presence of histocompatibility antigens on choriocarcinoma cells. By this method, normal villi, hydatidiform moles, and two choriocarcinomas did not have any apparent class I and II antigens. Differences in responsiveness to class I and II antigens were concisely reviewed by Nepom (1989).
When ABO blood group antigens were studied in a Lebanese popu- lation, families with trophoblastic disease had a normal distribution, as controls (Iliya et al., 1969). Bagshawe et al. (1971) observed that blood group A women married to blood group O husbands have the highest risk of trophoblastic neoplasia; those married to blood group A husbands have the lowest risk. Blood group AB patients have tumors that metastasized widely and were least affected by chemotherapy. The investigators considered that these effects show immunologic modula- tion of the tumor.
Because Rh(D) antigen is said to be expressed on trophoblastic sur- faces, Tomoda et al. (1981) suggested that hydatidiform moles [with Rh(D)] would be rejected by the Rh(d) host (see also Fischer et al., 1985), and that the difference in Rh antigens of Orientals could explain their high incidence of trophoblastic disease. Support for this idea has also come from the early studies by Scott (1962, 1968).
It is not yet certain that biologic trophoblast-specific antigen expres- sion exists. Cheng and Johnson (1988) endeavored to produce antibod- ies to such possible antigens. When the antibodies were carefully absorbed, these authors found a low incidence (3.6%) of corresponding antibodies in serum samples of infertile women. These investigators suggested that the earlier results of Grimmer and colleagues (1988) were erroneous, those that suggested a possible relation of these anti- bodies to lupus-like disorders. Srivannaboon (1971) found, in immuno- logic studies with a trophoblast line, that choriocarcinoma contained an antigen that was not shared by normal trophoblast. The nature of this antigen was not identified.
There have also been attempts to treat gestational trophoblastic neo- plasia with immunologic methods. Doniach et al. (1958) immunized a patient with choriocarcinoma by injecting her husband’s cells. Hackett and Beech (1961) injected concentrates of the husband’s leukocytes intradermally without inciting a reaction. They also employed Freund’s adjuvant therapy and noted a gradual diminution in metastatic tumor nodules. Final tumor resolution, however, occurred only after chemo- therapy. The conclusion of these authors was that choriocarcinoma is not antigenic in the usual sense. Others have not agreed with this con- clusion. Cinader et al. (1961) thought that their patient was successfully treated with leukocytes but were unable to rule out spontaneous remis- sion. Although some spontaneous remissions of choriocarcinoma are well documented (Brewer et al., 1961), this is not a common event and certainly spontaneous remission cannot be reliably anticipated in clini- cal management. Ewing (1941), in fact, did not accept such lesions that regressed as having been true choriocarcinomas.
The immunologic tolerance of the husband’s skin graft was explored in two patients with choriocarcinoma by Robinson et al. (1963), and Mathé et al. (1964) found variable skin graft survival in some patients with choriocarcinoma. There is no doubt that some GTNs elicit a cellular and immunologic response and that this may be related to survival (see Ober, 1969). It is also the presumed explanation for the poorer response to chemotherapeutic agents of the testicular choriocarcinomas in men because they lack foreign antigens. Until the precise mechanism of graft rejection is understood and we know more about the tolerance afforded to normal placental trophoblast, it is useless to engage in further specu- lation. The reader is referred to the numerous papers from symposia that have dealt with this question (Edwards et al., 1975; Beer &
Billingham, 1976; Wegmann & Gill, 1983; Gill et al., 1987).
EPIDEMIOLOGY OF CHORIOCARCINOMA
As with hydatidiform moles, the frequency of choriocarcinoma is irregu- larly distributed in different populations. It is also difficult to enumerate precisely because malignant moles are often included with choriocar- cinoma. Nevertheless, there are major differences in prevalence. For instance, Ho et al. (1989) found the prevalence of GTN to be lowest in whites (3/100,000 to 6/100,000) and highest in Chinese (68/100,000 to 202/100,000). They studied cellular antigens and found that Chinese had significant HLA sharing; whites did not. This finding suggested to them that genetic and nongenetic factors are of etiologic importance, and that this may also explain the generally better prognosis in whites. Rolon and Lopez (1979) summarized much of the epidemiologic aspects; their data and those of others are summarized in Table 23.3.
The higher incidence of choriocarcinoma in older women is also apparent in most of these studies. Oettle (1961) made the point that, despite the frequency of choriocarcinoma in Bantu women, there is no increased frequency of hydatidiform moles in that population. They believed that the higher incidence of choriocarcinoma in that popula- tion may result from higher ages at childbearing. In a study of Lebanese patients, Iliya et al. (1967) hypothesized that consanguinity is an important cause of choriocarcinoma. Parazzini et al. (1988a) adduced some data in favor of specific dietary deficiencies (e.g., vitamin A) in gestational trophoblastic neoplasia; “however, the limitation of avail- able evidence still introduces serious uncertainties in the interpretation of these findings.”
Endocrine Aspects of Gestational Trophoblastic Neoplasia
The most important endocrine consideration of chorio- carcinoma is its production of hCG or subunits thereof (see also Chapter 22). Follow-up of patients with CHM and choriocarcinomas is effectively done by assay of this hormone in serum (Delfs, 1959). Urine hCG values above 100,000 IU/24 hours (or serum levels of b-subunits measured by radioimmunoassay) strongly suggest the presence of a hydatidiform mole or malignant GTN.
Conventionally, patients are monitored with weekly
determination of hCG levels until none is detected for 3
consecutive weeks after the initial treatment of complete
moles. In the usual case, hCG becomes undetectable
within 8 to 173 days (Yuen et al., 1977). Chorionic gonad-
otropin is cleared by renal mechanisms, with a half-life of
35 hours (Midgley & Jaffe, 1968). Hammond et al. (1981)
suggested that, after the titers have disappeared, hCG
should be monitored for 1 year after evacuation of a
mole. Even so, late recurrences have been reported and
they have frequently occurred despite adherence to this
regimen (e.g., Kirk, 1965; Vaughn et al., 1980). The obser-
vations by Rotmensch and Cole (2000) in the treatment
of possible GTNs on the basis of elevated hCG titers
should be appreciated. These authors showed clearly that
elevated hCG titers, by themselves, are not a reliable indi-
cation of trophoblastic tumors. This is due to the variable
nature of antibodies against hCG used commercially and
perhaps also to the possible binding of preexisting anti-
bodies to mouse immunoglobulin G (IgG). Inasmuch as
this elevation of titers had led to the inappropriate therapy
of 12 patients, they recommended greater scrutiny in the face of such laboratory findings. A possible improvement of monitoring cases with GTN is perhaps feasible with the determination of b-subunits of hCG in urine by radio- immunoassay (Wehmann et al., 1981). Through simulta- neous measurements of serum and spinal fluid b-hCG levels, studies suggested that it may be possible to anti- cipate cerebral metastases (Soma et al., 1980).
Yazaki et al. (1980) compared isoelectric homogeneity of serum hCG in various types of normal and abnormal pregnancies. They found that normal gestations, CHM, and invasive moles had “normal” isoelectric peaks, whereas additional peaks were detected in sera of patients with choriocarcinoma. A variety of studies have shown that the hCG originates from the syncytial component of the choriocarcinoma. When Dawood (1975) measured progesterone levels in patients with choriocarcinoma, he found that it correlated with hCG. He believed that the elevated progesterone levels resulted from its production by the hCG-stimulated ovaries, as many patients experi- ence theca lutein cysts with this neoplasm (Kohorn, 1983), rather than the trophoblast. Lacking appropriate fetal adrenal precursors, estrogen production is sparse in cho- riocarcinoma patients, and its determination is not useful.
It is different in men with choriocarcinoma. They have excessive estrogen production and often suffer gyneco- mastia (Martin & Carden, 1963), presumably because of Leydig cell stimulation. Follow-up with hPL, MBP, and other trophoblastic proteins has either not been under- taken or has not been useful for therapeutic and management purposes.
Occasionally, choriocarcinoma is associated with exces- sive androgen production. These rare reports have been competently reviewed by Nagelberg and Rosen (1985).
The origin of the androgens is still in dispute, although some in vitro work suggested a trophoblast origin. Searle et al. (1978) measured a placenta-specific b
1-glycoprotein in the serum of patients with GTN and found a moderate elevation in CHM and choriocarcinoma patients—much
less than that of hCG, however. They were not enthusias- tic about the use of this marker in the management of patients. Nisbet et al. (1982) localized another placenta protein marker (PP5) to syncytium of placentas and its tumors. Serum levels, however, were not sufficiently ele- vated for its clinical use.
PSTTs are also associated with increased production of hormones, specifically hCG. Serum levels of hCG, however, are only moderately elevated, usually up to only 10,000 IU, as opposed to the extremely high levels seen in choriocarcinoma. In addition, only about 75% to 80%
of patients show any elevation at all (Baergen & Rutgers, 1997).
Ectopic Choriocarcinomas; Tumors in Men
Aside from the truly gestational choriocarcinoma that most commonly follows a complete hydatidiform mole, choriocarcinoma has occasionally been reported in the fallopian tube, ovary, and other locations. These lesions may then be referred to as heterotopic chorioepithelioma (Nanke, 1959). Nanke, however, discovered such a tumor after hysterectomy was performed for hydatidiform mole, and it is most likely that the CHM was the tumor’s ante- cedent of the cerebral, renal, and pulmonary metastases.
The author opined that the tumor arose from deportation of benign trophoblast that subsequently dedifferentiated.
He termed it “chorioepitheliosis.” As detailed in Chapter 22, such uncertainties can now be resolved when genetic markers are studied. Three rare cases of choriocarcinoma in tubal pregnancies were the topic of a publication by Horn et al. (1994). All were successfully treated and metastases did not occur.
The primary choriocarcinoma in the lung of a patient without any other lesions, reported by Tanimura et al.
(1985), most likely resulted from a prior therapeutic abortion. To support such a proposal, the authors exam- ined the lungs of 10 patients dying after delivery or abor- tion and found pulmonary trophoblast in nine. Tubal
Table 23.3. Frequency of choriocarcinoma and chorioadenoma destruens among number of pregnanciesAuthors Region Choriocarcinoma Chorioadenoma destruens
Hertig & Mansell (1956) US 43,000
Wei & Cuyang (1963) India 912
Mills (1964) England 70,000
Tow (1965) Singapore 5,000
Acosta-Sison (1966) Philippines 1,382–2,229
Kolstad & Hognestad (1965) Norway 20,000
Aranda & Martinez (1968) Puerto Rico 32,460
Yen & MacMahon (1968) Rhode Island 40,000
Reddy & Rao (1969) Taiwan 496
Mogensen & Olsen (1972) Denmark 49,000 126,000
Matalon et al. (1972) Israel 15,000
Baltazar (1976) Philippines 5,747
Rolon & Lopez (1979) Paraguay 43,489 70,252
Source: Modified from Rolon & Lopez (1979).
choriocarcinoma was reported by Madden (1950) follow- ing tubal abortion, as did most of the 48 other cases reviewed by the author. Other reports on tubal chorio- carcinoma have come from Riggs et al. (1964) and Lurain et al. (1986). A comprehensive review was published by Ober and Maier (1981), who found 93 cases in the entire literature, deemed 58 to be acceptable, and added 18 of their own. These investigators indicated that small tumors are difficult to distinguish from ectopic pregnancy. They found two cases that still contained villi and thus should not be called choriocarcinoma. Some patients were cured by simple hysterectomy, whereas others (94%) responded well to chemotherapy.
Ectopic lesions of extravillous trophoblast have rarely been reported. Baergen et al. (2003) added seven new cases to the previous 12 descriptions in the literature of extrauterine lesions of intermediate trophoblastic prolif- eration. Follow-up was benign and an origin from ectopic pregnancies was postulated. The fact that trophoblast is so highly susceptible to methotrexate has led to many trials of the medical treatment of ectopic pregnancies with this agent (Ory et al., 1986; Carson et al., 1989). Five of six patients had no further need for therapy; one required subsequent salpingectomy. As described earlier, Barnes et al. (1982) reported the successful treatment of a placental choriocarcinoma, metastatic to the brain and lung. Following two normal pregnancies, however, this was followed by a primary lethal brain glioblastoma 7½ years after the choriocarcinoma, a lesion that was perhaps secondary to irradiation. Creinin et al. (1998) studied the action of methotrexate in early pregnancy and concluded that its main result is the disruption of the change from cytotrophoblast to syncytium—cytotrophoblast syncytial- ization as they called it.
Some choriocarcinomas occur without known anteced- ent pregnancy (Acosta-Sison, 1957). Benjamin and Rorat (1978) reported ovarian choriocarcinomas as “primary.”
Their patient developed the fatal ovarian (and hepatic) tumor apparently spontaneously. The uterine cavity con- tained only necrotic debris. The authors found six previ- ous cases similar to theirs in the literature. Turner et al.
(1964) reported on a successfully treated patient with an apparently primary ovarian choriocarcinoma. No ante- cedent or coincident pregnancy was found, although the authors called the tumor gestational. Cunanan et al. (1980) reported an ovarian choriocarcinoma that was associated with a normal uterine pregnancy delivered by hysterot- omy several weeks after resection of the mass. The patient did well on chemotherapy; the infant and placenta were normal. This report included a review of 24 additional choriocarcinomas coexisting with pregnancy. In 10, the tumor was found in the placenta, in five it was absent, and in the remaining 10 cases there was a lack of information.
Axe et al. (1985) found six similar cases in their Registry of Ovarian Tumors. One of those patients died.
Manivel et al. (1987) reviewed the extensive literature of trophoblastic differentiation from germ cell tumors.
This tumor is outside the scope of the present book.
Suffice it to say that these investigators paid special atten- tion to the presence of intermediate trophoblast in these germinal tumors. The cells were designated large mono- nuclear cells with a variety of immunologic characteris- tics that had led Kurman et al. (1984) to call them intermediate trophoblast. They should now be studied with more specific markers, such as MBP, and the designa- tion “extravillous trophoblast” is more appropriate. Cho- riocarcinoma may differentiate from primary germ cell tumors of testis and mediastinum, and from teratocarci- nomas. Fine et al. (1962) reviewed this topic adequately and reported a fatal case in which the tumor arose in the mediastinum, and paid special attention to the endocrine aspects of the neoplasm.
Wenger et al. (1968) reported a mediastinal chorio- carcinoma in a man. This tumor is unusual. They had seen only 17 choriocarcinomas in men at the Mayo Clinic. Chemotherapy is significantly less effective in men than for trophoblastic neoplasms, which has been assumed to be due to the assistance of immunologic rejection mechanisms in the gestational tumor. Soma et al. (1973), who studied the chromosomes of a man with gastric choriocarcinoma, identified fluorescent Y bodies in this hyperdiploid cell line. Fukunaga and Ushigome (1993a) found their eight tumors to be diploid and reviewed the evidence that suggested that chromosomes of these tumors are normally diploid. A primary chorio- carcinoma without antecedent lesion was described in the lung of a 51-year-old man (Sullivan, 1989). A pul- monary choriocarcinoma was also reported in a 60-year- old woman (15 years after menopause). Because of her age, it is unlikely that it had derived from a preceding gestation (Pushchak & Farhi, 1987). Pushchak and Farhi speculated that it came from epithelial metaplasia, as she (and the patient of Sullivan) had been heavy smokers.
These cases must all be interpreted with caution, however, especially when one knows of the remarkable patient described by Dougherty et al. (1978): an 86-year- old woman with fatal pulmonary and cerebral metastases whose uterus had choriocarcinoma with degenerated villi. The inference was that the villi and trophoblast had existed since her last pregnancy, 34 years earlier. She had been amenorrheic after that pregnancy. This case is truly exceptional. Another remarkable case is that reported by Lathrop et al. (1978) of a choriocarcinoma that devel- oped 14 years after tubal ligation.
THERAPY OF GESTATIONAL TROPHOBLASTIC NEOPLASIA
Aside from the attempts at immunotherapy discussed previously, surgery and chemotherapy are used for the treatment of choriocarcinoma and its precursors. Initially, only surgical excision, perhaps prophylactic
