Int J Clin Oncol (2000) 5:71–78
REVIEW ARTICLE
Harold Fox
Trophoblastic disease
Received: February 16, 2000
Key words Trophoblast · Mole · Choriocarcinoma
Introduction
The term “gestational trophoblastic disease” is, by conven-
tion, restricted to hydatidiform moles, choriocarcinoma,
and placental site trophoblastic tumor. These are all defined
in purely morphological terms, but any classification of
trophoblastic disease also includes one non-morphological
component, namely, “persistent trophoblastic disease”; this
term is applied to a biochemical abnormality, ie, an elevated
level of human chorionic gonadotrophin (hCG) following
a molar pregnancy, and this diagnosis not only lacks any
specific morphological connotation but becomes invalid if a
morphological diagnosis is achieved.
The term “gestational trophoblastic neoplasia” is some-
times used as an alternative to “gestational trophoblastic
disease”, and, indeed, many seem to feel, often at a sublimi-
nal level, that these various conditions represent a neoplas-
tic spectrum, with moles at the benign end of this spectrum,
choriocarcinoma at the malignant extreme, and invasive
hydatidiform mole being equivalent to a neoplasm of bor-
derline malignancy. This is a totally misleading approach,
for there is nothing to suggest that a hydatidiform mole, of
any type, is a form of neoplasia; it is, without question, a
specific form of abnormal pregnancy. A choriocarcinoma is
H. Fox (*)
Department of Pathological Sciences, The University of Manchester,
Stopford Building, The University of Manchester, Oxford Road,
Manchester M13 9PT, England
Tel. 144-161-275-5301; Fax 144-161-275-5289
This paper was partly presented at The 10th Anniversary Annual Re-
view Course on Gynecologic Oncology and Pathology in Otsu, Japan,
October 25–29, 1999.
© The Japan Society of Clinical Oncology 2000
usually considered to be neoplastic, but, as will be discussed
later, even this may be no more than an unusual type of
abortus in some patients, and the only undoubtedly neo-
plastic entity within this group of conditions is the placental
site trophoblastic tumor.
Hydatidiform moles
During the last few decades molar disease has been divided
into complete and partial hydatidiform moles, the distinc-
tion between these two entities being originally based on
their differing morphological features.1,2
Morphology
Classically, a complete mole is characterized by diffuse ve-
sicular change, which involves, to a greater or lesser degree,
the entire villous population, thus producing the classical
“bunch of grapes” appearance; the mass of villous tissue is
markedly increased, to an extent that an in-situ mole can fill
or even distend the uterus, with no remnant of a normal
placental shape being discernible and neither a fetus nor a
gestational sac being present. In a complete hydatidiform
mole there is generalized vesicular distension of the villi,
the entire villous population being involved to a greater or
lesser extent; the vesicular villi are usually evident to the
naked eye in material retrieved by curettage but tend to be
collapsed, and not grossly visible, in tissue obtained by suc-
tion. Histologically, the distension, which is of variable de-
gree, of the villi is seen to be caused by an accumulation of
fluid in the villous stroma, and there is, commonly, central
cavitation (cistern formation). It is usually maintained that
fetal tissues are not found in association with a complete
mole, but, while villous stromal vessels are usually absent,
attenuated vestiges of such vessels are apparent in a small
minority of cases and in occasional complete moles there
may be quite well formed fetal vessels containing nucleated
red blood cells; amniotic tissue may also occasionally be
72
seen.3 A characteristic feature is the presence of stromal
nuclear debris.4,5 All moles show, by definition, some degree
of atypical trophoblastic proliferation, although this is not
necessarily present in all the villi, some of which may have
a flattened trophoblastic mantle. Most villi do, however,
show trophoblastic proliferation, and this may be circumfer-
ential, focal, or multifocal, and of slight, moderate, or
marked degree. It should be noted that trophoblastic prolif-
eration as marked, or even greater, than that encountered
in a mole may be seen in villi from a normal first trimester
pregnancy. In the normal placenta, however, the proliferat-
ing trophoblast is always at one pole, or along one side, of
the villus and never shows the focal or circumferential
pattern characteristic of a mole. Variable nuclear atypia is
invariably present in the proliferating trophoblast, although
this is commonly no more marked than the atypia that is
often seen in normal first trimester placentas. There are no
morphological features which allow for a histological dis-
tinction to be made between diploid and tetraploid com-
plete moles.6
This classical description of complete hydatidiform
moles is based on cases that have been diagnosed clinically.
With the increasing use of ultrasound, complete moles are,
however, now being diagnosed at an earlier stage of gesta-
tion than in the past. In these early lesions, vesicular villi are
often not visible to the naked eye, and, histologically, there
may be co-existing vesicular and non-vesicular villi; those
villi that are not vesicular are often branching and have a
polypoid or lobulated appearance.4 Fetal vessels will be
present in many of the villi, and these are set in a somewhat
mucoid stroma in which there are a few spindly cells; stro-
mal nuclear debris is present in the villi of an early complete
mole and is a defining feature.
A partial mole differs in numerous respects from a com-
plete mole: it is commonly associated with a fetus, retains a
placental shape, and does not have a greatly increased vil-
lous mass. Only a proportion of the villi show macroscopi-
cally visible vesicular change, these villi being scattered
within normal placental tissue. Histologically, the presence
of both edematous swollen villi and vascularized villi of
normal size, albeit often with an unusually fibrotic or cellu-
lar stroma, is confirmed. The vesicular villi commonly have
a very irregular, scalloped outline, this resulting in the
“Norwegian fjord” appearance; the cutting of some of these
deep indentations in cross section results in the presence
of so-called “trophoblastic inclusions” within the villous
stroma. There is abnormal trophoblastic proliferation, and
this, as with complete moles, is usually either circumferen-
tial or multifocal; the degree of trophoblastic proliferation
is usually not only less than that seen in complete moles but
is less than that encountered in normal first trimester pla-
centas. The villous trophoblast frequently has a somewhat
vacuolated, or “lacy”, appearance. It is often maintained
that partial moles are characterized by “syncytiotropho-
blastic hyperplasia”, but it is, of course, quite impossible for
the villous syncytiotrophoblast to be hyperplastic, for this is
a post-mitotic, terminally differentiated tissue which is inca-
pable of DNA synthesis and cell division. In some partial
moles an angiomatoid appearance is seen.
It should be noted that the use of the word “partial” to
describe a mole does not mean that one portion of the
placenta is normal and another portion molar. This latter
situation is encountered if a complete mole, or very much
less frequently a partial mole, is part of a dizygotic twin
pregnancy and associated with a normal twin;7 in such preg-
nancies one placenta is molar and the other non-molar,
while in a true partial mole there is an intermingling of
molar and non-molar villi, with the molar villi distributed
throughout the entire placenta.
Cytogenetics
It has been recognized for some time that complete moles
are androgenetic, ie, all their nuclear DNA is paternally
derived; furthermore, approximately 95% of complete
moles have a 46XX chromosomal constitution, the remain-
der having a 46XY karyotype. Banding studies have shown
that the vast majority of 46XX complete moles are derived
from a single sperm (monospermic or homozygous moles),
while all 46XY moles and a small minority of 46XX moles
are derived from two sperms (dispermic or heterozygous
moles). A hypothetical model has been proposed to explain
these findings; namely, that homozygous moles are caused
by fertilization of a “dead” ovum (ie, one containing no
viable genomic material) by a single haploid sperm, which
then undergoes endoreduplication of its genetic material
without cell division, and that heterozygous moles result
from entry into a “dead” ovum of two haploid sperms,
which then fuse and replicate. It has to be stressed that this
concept is purely theoretical and that the mechanism of the
loss of the maternal genome is unknown; there may be a
primary loss of the genome, or maternal chromosomes may
be excluded in the cell divisions following fertilization. That
fertilization has actually occurred in molar gestations is cer-
tain, for the mitochondrial DNA is of maternal origin.
Partial moles contain both paternal and maternal ge-
nomic material, and the vast majority are chromosomally
triploid, usually 69XXY but sometimes 69XXX or 69XYY.
It is, of course, known that not all triploid gestations are
associated with molar change in the placenta, and it is now
clear that if the extra chromosomal load is of paternal origin
a partial mole will result, while if the extra chromosomal
content is contributed by the mother a non-molar placenta
will develop.8 It is thought that paternally derived triploidy
is usually the result of two haploid sperms fertilizing a hap-
loid ovum, but that a few cases may be caused by fertiliza-
tion of a haploid ovum by a diploid 46XX sperm.
This relatively simple subdivision into androgenetic
diploid complete moles and biparentally derived triploid
partial moles is not, however, the whole story. Cytogenetic
studies have yielded a few instances of triploid and tetrap-
loid complete moles and of diploid and tetraploid partial
moles.9 Further, flow cytometric studies, while in general
confirming that most complete moles are diploid and most
partial moles triploid, have, nevertheless, rather consis-
tently revealed small subpopulations of diploid partial
moles and triploid complete moles.10 To complicate the

matter still further, there have been very occasional in-
stances of androgenetic partial moles and biparently de-
rived complete moles,9 entities for which there is, at the
moment, no very plausible explanation.
It should be noted that these anomalies have been de-
tected in moles for which a rigorous morphological diagno-
sis had been made. Triploid complete moles may well be
caused by three haploid sperms entering a “dead” ovum,
but many of the reported diploid partial moles have, in
reality, been examples of a complete mole with an accompa-
nying concomitant non-molar twin pregnancy; nevertheless,
a distinct entity of diploid partial mole does exist, which, it
has been suggested, merits consideration as a separate third
type of molar pregnancy, such cases possibly being a result
of uniparental disomy or of malfunction in the maintenance
of imprinting.11
Studies of mice models12 have indicated that paternal
genes play a dominant role in placental development, while
maternal genes have a major role in fetal development,
this being the result of genomic imprinting. This concept
would be in accord with the cytogenetic findings in com-
plete moles, which, with their double content of paternal
alleles, might express a double dose of a gene such as IGF2,
which is genomically imprinted to the paternal copy. How-
ever, the maternally imprinted gene H19 is also expressed in
complete moles,13 and this suggests that imprinting is lost,
possibly because the imprinting process may require a bipa-
rental genome.
Differential diagnosis of molar pregnancy
Because all women with molar disease are subjected to
follow-up, the distinction, in biopsy material, between
partial and complete moles is a useful, but not a vital, one
to make. In practice it is relatively easy to distinguish, on
morphological grounds alone, between a classical complete
mole and a partial mole. Difficulties do arise, however,
when an early complete mole is encountered, for, under
these circumstances, neither the finding of villous fetal ves-
sels nor an admixture of vesicular and non-vesicular villi is
of any distinguishing value; greater stress has to be placed
on the lobulated or polypoid appearance of the villi and the
presence of stromal nuclear debris in complete moles.4 If it
is felt that a precise diagnosis is required in truly doubtful
cases of molar disease, this can usually be achieved with
flow cytometry, the complete moles having, with rare excep-
tions, a diploid or tetraploid DNA content and the partial
moles a triploid DNA content.14
The distinction between a hydatidiform mole and a hy-
dropic abortion is one of considerable practical importance,
for on this rests the decision as to whether a woman requires
surveillance or not. It is usually quite easy to distinguish
between a complete mole and a hydropic abortion, al-
though this is not invariably the case,15 but serious difficul-
ties may be encountered in making a distinction between a
partial mole and a hydropic abortus. The criterion for mak-
ing this distinction is the contrast between the highly irregu-
lar vesicular villi with central cistern formation and the focal
73
trophoblastic hyperplasia seen in the partial mole and the
rounded villi, lacking cisternal change, and showing no
abnormal trophoblastic proliferation, seen in a hydropic
abortion. It has to be admitted, however, that while on
paper it appears relatively easy to distinguish a partial mole
from a hydropic abortus, it is unfortunately true that in
practice there are very considerable interobservor differ-
ences in making this differential diagnosis.16 Flow cytometry
is of some help, but it has to be borne in mind that some
non-molar hydropic abortions also have a triploid DNA
content.
An uncommon, but difficult, diagnostic problem may
occur if curettage material is received from a twin preg-
nancy in which one twin has a normal placenta and the
other is a complete hydatidiform mole, the difficulty being
that the admixture of vesicular and non-vesicular villi may
suggest a diagnosis of a partial mole. Under such circum-
stances, it may be possible to discern that the vesicular villi
have the features of a complete rather than a partial mole,
but an absolute diagnosis depends upon showing that all the
villi are diploid and that, while some are biparental, others
are androgenetic.18
Postmolar disease
In the United Kingdom, about 8% of patients who have had
a complete mole will develop persistent trophoblastic dis-
ease; the figure in the United States is rather higher because
of the use of different diagnostic criteria. The actual cause
of persistent trophoblastic disease is generally unknown.
There may have been incomplete removal of the mole, but
it is equally possible that many of these cases are invasive
hydatidiform moles with residual invasive molar tissue
within the myometrium or its vasculature; it is also possible
that some are early cases of choriocarcinoma. It is thought
that the risk of development of a clinically overt choriocar-
cinoma following a complete mole is less than 5%.
The incidence of persistent trophoblastic disease follow-
ing a partial mole has been much disputed, but there is now
no doubt that it occurs, although the magnitude of the risk
is very much lower than that for complete moles. The even-
tual risk of choriocarcinoma in patients with partial moles is
also unknown; choriocarcinomas have been reported fol-
lowing partial moles, but this, in itself, does not necessarily
mean that this condition increases the risk of a choriocarci-
noma, for choriocarcinoma can follow a normal pregnancy.
Prognostic factors
As it is widely agreed that all women who have had a molar
pregnancy should enter a follow-up surveillance program,
there is, therefore, no practical point in attempting to define
those women at most risk of developing postmolar disease.
Nevertheless, from a purely theoretical viewpoint, it is of
interest to consider whether a high-risk group can be de-
fined. It was at one time considered that the risk of eventual
postmolar disease was directly related to the degree of tro-
phoblastic proliferation in the mole, but this has proved not
74
to be the case,19 and the “grading” of moles in terms of their
degree of trophoblastic proliferation has now been aban-
doned. Attempts to identify those women at greatest risk
for postmolar disease by the use of cell proliferation mark-
ers, the expression of proto-oncogenes such as c-erb-B2,
and flow cytometry20–22 have failed, but it has been main-
tained that heterozygous (dispermic) moles have a much
higher risk of subsequent postmolar complications than do
homozygous (monospermic) moles.23 Doubts have, how-
ever, been cast upon this claim by the failure to find any
association between the presence of a Y chromosome (de-
tected by the polymerase chain reaction or by chromosome
in situ hybridization) in a mole and an excess incidence of
postmolar disease.24,25
Invasive hydatidiform mole
An invasive hydatidiform mole is one which penetrates
into the myometrium or invades the uterine vasculature. A
deeply invasive mole usually becomes clinically evident sev-
eral weeks after apparently complete evacuation of a mole
from the uterus, the patient usually presenting with hemor-
rhage. If a hysterectomy is performed at this stage, the
appearances range from, at one extreme, only a small hem-
orrhagic focus in the myometrium to, at the other end of the
spectrum, a large deeply cavitating hemorrhagic lesion of
the uterine wall which mimics a choriocarcinoma. Rarely, a
mole penetrates the full thickness of the myometrium, this
leading either to uterine perforation or to extension of the
mole into adjacent structures, such as the broad ligament.
The histological distinction from a choriocarcinoma is de-
pendent upon the finding of molar villi within the uterine
wall, these more commonly being seen in the myometrial
vascular channels than between the myometrial fibers; the
molar villi show a very variable degree of trophoblastic
proliferation and sometimes this is far from being a con-
spicuous feature.
Invasive moles have, in the past, caused death from uter-
ine bleeding or perforation, but their mortality rate is now
virtually zero because of the success achieved in their treat-
ment by a limited course of chemotherapy. In fact, the
diagnosis of an invasive mole, which can be made with
certainty only in a hysterectomy specimen, is now almost
obsolete because nearly all invasive moles are subsumed
into the category of persistent trophoblastic disease.
It has to be stressed the the invasive capacity of some
moles is not an indication that they are neoplastic. Normal
trophoblast has the ability to invade both the myometrium
and the uterine vessels, while villi from a normal placenta
can invade deeply into, or even through, the uterine wall to
give rise to a placenta increta or a placenta percreta, both of
which are the exact non-molar equivalents of an invasive
mole. Molar tissue can be transported via the bloodstream
to extrauterine sites, particularly to the vagina and lungs;
the transported molar trophoblast can then grow in these
sites to form nodules that are either clinically or radiologi-
cally detectable. The development of “metastatic” lesions
implies that molar trophoblast has entered the uterine ves-
sels and, hence, their presence is taken as de-facto evidence
of the presence of an invasive mole; the “metastatic” nod-
ules are not, however, usually associated with evidence of
molar invasion of the myometrial tissues.
The “metastatic” nodules commonly appear several
weeks after evacuation of a mole from the uterus, but they
may occur concurrently with a mole or can be the present-
ing symptom of such a lesion. Vaginal lesions form he-
morrhagic submucous nodules; histologically, these often
contain villous structures, a finding which rules out a diag-
nosis of choriocarcinoma. Even if villi are not present in a
postmolar vaginal lesion and the mass is formed of biphasic
trophoblast, a diagnosis of choriocarcinoma is not war-
ranted if the trophoblast is not invading normal tissues;
following a non-molar pregnancy, the diagnosis of a lesion
such as this would be choriocarcinoma, even in the absence
of tissue invasion.Pulmonary lesions can cause hemoptysis
but are usually an asymptomatic radiological finding; histo-
logical examination of these will also usually reveal the
presence of villi. These extrauterine lesions may resolve
spontaneously but are commonly treated with limited che-
motherapy, which achieves excellent results.
The fact that molar trophoblast is transported to extrau-
terine sites is not an indication of neoplastic behavior. The
trophoblast enters the maternal bloodstream in every nor-
mal pregnancy,26 and is transported to sites such as the
lung;27 this transported trophoblast only gives rise to detect-
able lesions if it is molar in nature.
Choriocarcinoma
Approximately 50% of choriocarcinomas follow a molar
pregnancy, 30% occur after an abortion, and 20% follow
an apparently normal gestation. The time interval between
the antecedent pregnancy and the clinical presentation of a
choriocarcinoma is very variable, ranging from a few weeks
or months to 15 years.
Morphology
Within the uterus a choriocarcinoma forms single or mul-
tiple hemorrhagic nodules which are often accompanied by
local metastases to the cervix and vagina. The neoplastic
masses consist of a central area of hemorrhagic necrosis
and, usually, although not invariably, a peripheral rim of
viable tumor tissue. The central, sometimes complete, ne-
crosis of the neoplastic tissue is a reflection of the fact that
a choriocarcinoma has no intrinsic blood supply, relying
for its oxygenation on its ability to invade the uterine blood
vessels; it is, therefore, only the growing edge of the tumor
which is adequately oxygenated, the remainder undergoing
ischemic necrosis.
Histologically, a choriocarcinoma typically has a biphasic
structure that recapitulates, often to a striking degree, that
of the trophoblaast of the normal implanting blastocyst,

central sheets or cores of cytotrophoblast being “capped”
by a peripheral rim of syncytiotrophoblast. The trophoblas-
tic cells in a choriocarcinoma commonly show no greater
degree of atypia and mitotic activity than that seen in an
implanting blastocyst. Villi are never present in an extra-
placental choriocarcinoma and, indeed, the presence of
villous structures negates a diagnosis of choriocarcinoma.
Although a biphasic structure is characteristic of choriocar-
cinoma, a monophasic variant, composed solely of cytotro-
phoblast, has been anecdotally described.
Because of the need to obtain an oxygen supply, a cho-
riocarcinoma is avariciously invasive of vascular channels in
the myometrium, in vessels which, it should be noted, are
also invaded by trophoblast during the process of normal
implantation. The tumor cells tend to form solid plugs
within the myometrial vasculature and, although there is
often extravascular extension, the malignant trophoblast
tends to infiltrate between the muscle fibers with very little
tissue destruction. The propensity for vascular invasion is
the basis for the predominantly hematogenous dissemina-
tion of a choriocarcinoma to sites such as the lungs, brain,
liver, kidney, and gastrointestinal tract; large tumor emboli
may impact within the pulmonary arteries. Lymph node
deposits of a choriocarcinoma are usually tertiary me-
tastases from a large extrauterine lesion.
Origin
There are many puzzling aspects of choriocarcinomas, such
as their status as an allograft (which they must be because
of their content of paternal antigens), their increased fre-
quency in women of blood groups A with group O spouses
and in group O women with group A spouses,28 and their
rather strange, almost bizarre, epidemiological risk fac-
tors, which include dieting, a family history of dizygotic
twins, more than one marriage, and infrequent sexual inter-
course;29 by far the most perplexing problem they pose is,
however, their origin. As already remarked, choriocarcino-
mas may follow either a normal or a molar pregnancy, more
commonly the latter, and there is a considerable gap in
our knowledge as to the relationship between the previous
pregnancy and the subsequent choriocarcinoma. Is the neo-
plasm actually derived from the trophoblast of the prior
pregnancy and, if so, what has been happening to this
trophoblast during the intervening months or years? The
application of genetic techniques, such as the study of cyto-
genetic polymorphism,30 DNA restriction fragment-length
polymorphism assays,31,32 or the study of tandem repeat re-
gions amplified by the polymerase chain reaction,33,34 has
shown that some, but by no means all, choriocarcinomas are
androgenetic; a significant proportion are, however, bipa-
rental.35,36 It has been presumed that the androgenetic tu-
mors are derived from a previous molar gestation, and this
has indeed been proven to be so in several patients, by the
showing of complete genetic identity between the mole and
the choriocarcinoma, this despite the fact that in two such
patients a full-term pregnancy with normal delivery had
intervened between the molar pregnancy and the develop-
75
ment of the choriocarcinoma.33,34,37 In one of these patients,
the time interval between the mole and the choriocarci-
noma was 10 years, and it is difficult to understand how
tissue from the mole remained in the uterus for that length
of time, and throughout a later normal pregnancy, to then
subsequently undergo a neoplastic resurgence, this typify-
ing the questions posed by this enigmatic lesion.
Some points are, however, becoming clearer; it seems
increasingly probable that many, possibly most or even all,
choriocarcinomas that follow an apparently normal preg-
nancy are, in reality, metastases from an undetected small
intraplacental choriocarcinoma. Only a small number of
intraplacental choriocarcinomas have been described, and
in nearly all the choriocarcinoma was very small and easily
overlooked unless the placenta was meticulously examined;
in most cases the placenta had been subjected to such ex-
amination because metastases had developed in the mother
during pregnancy; in only two cases had the tumor been
detected in the absence of such complications.38,39 One
case40 is of particular interest, in so far as a patient devel-
oped an apparently primary choriocarcinoma soon after
a normal pregnancy; re-examination of the placenta at
that time revealed a tiny intraplacental choriocarcinoma.
Intraplacental choriocarcinomas are histologically identical
to extraplacental choriocarcinomas, but are often separated
from the normal villous population by villi with a surround-
ing mantle of choriocarcinoma-like trophoblast which has
replaced the normal trophoblast; this therefore confirms the
long-held opinion, based largely on the ability of choriocar-
cinomas to secrete hCG, that a choriocarcinoma is a lesion
of villous trophoblast, despite the invariable absence of villi
from extraplacental tumors. It is not yet known whether all
choriocarcinomas following a normal pregnancy are bipa-
rental rather than androgenetic, but, if this were proven to
be so, it would strengthen the case for such lesions being
metastases rather than primary neoplasms.
If choriocarcinomas following a normal pregnancy are,
in reality, metastases from an intraplacental choriocarci-
noma, are those which follow a molar gestation similarly
derived from an intramolar choriocarcinoma? This is cer-
tainly a possibility, for one such lesion has been described,41
although it was not associated with subsequent disease. The
prolonged time interval, in many patients, between a molar
pregnancy and the development of an overt chorio-
carcinoma does, however, suggest that by no means all
postmolar choriocarcinomas are derived from intramolar
lesions, and it is certainly conceivable that some such cho-
riocarcinomas are new pregnancies, the choriocarcinoma ab
initio that has long been considered as a possibility.42 Some
choriocarcinomas which have not obviously followed a
normal pregnancy have been biparental or even of purely
maternal origin, and there seems no good reason why a
pregnancy, of any genetic type, should not evolve directly
into a choriocarcinoma. There is an excellent precedent for
the belief that a pregnancy can appear to be neoplastic;
namely, the now generally agreed concept that a teratoma is
a parthogenetic pregnancy. This raises the question, how-
ever, as to whether such pregnancies should be considered
as truly neoplastic or simply as aberrant gestations. Chorio-
76
carcinomas invade vessels and spread to distant sites, but so
does normal trophoblast; histologically, choriocarcinomas
resemble acutely the trophoblast of the normal implanting
blastocyst, and their response to methotrexate, while being
quite different to that of virtually every other neoplasm,
is not unlike that of a normal, but ectopic, early gestation.
This is, of course, pure speculation but, nevertheless, the
possibility that some choriocarcinomas are simply abnormal
pregnancies should not be dismissed too lightly.
Placental site trophoblastic tumor
This is an uncommon tumor that is derived from the
extravillous trophoblast of the placental bed (sometimes
referred to as “intermediate trophoblast”). In the vast
majority of patients, the neoplasm develops after a normal
pregnancy, only 5% occurring after a molar gestation.43 Pa-
tients present at anything from a few weeks to 18 months
after the antecedent pregnancy with a complaint of irregu-
lar vaginal bleeding or, perhaps more commonly, amenor-
rhea; a small proportion of patients develop a nephrotic
syndrome which appears to be caused by chronic intravas-
cular coagulation initiated by factors released from the tu-
mor. The uterus is commonly enlarged, but a pregnancy test
is positive in only one-third of patients, this reflecting the
fact that the principal secretory product of extravillous
trophoblast is human placental lactogen (hPL) rather than
hCG.
The tumors tend to form tan, white, or yellow masses
within the myometrium and often protrude into the en-
dometrial cavity, sometimes forming a polypoidal mass.
Histologically, the tumor replicates, in an anarchic form, the
appearances seen in the normal placental bed. The tumor
is composed predominantly of mononuclear, polygonal,
cytotrophoblastic-like cells which infiltrate between and
dissect the myometrial fibers as cords or sheets; spindl-
ing of cells is sometimes seen. Coagulative necrosis and
hemorrhage are often present, but are sometimes notably
absent. Although the classical bimorphic pattern of a
choriocarcinoma is not seen, a number of irregularly
distributed multinucleated cells are usually present. The
cytotrophoblastic-like cells that predominate in these neo-
plasms tend to stain positively for hPL rather than hCG.
Fibrinoid necrosis of vessel walls and deposition of fibrinoid
material around tumor cells is very commonly present,
while tumor cells are present within the blood vessels in
most, but not all, patients, and entrapped vessels not un-
commonly show fibrinoid necrosis. The endometrium adja-
cent to the neoplastic cells may have a pseudodecidual
appearance. Most placental site trophoblastic tumors ap-
pear to be benign, but between 10% and 15% behave in a
malignant fashion.43
The diagnosis of a placental site trophoblastic tumor in
curettage material is far from easy,44 the major difficulty
being in attempting to distinguish between a neoplasm of
this type and an exaggerated, but non-neoplastic, placental
site reaction. One of the most useful criteria in making this
differential diagnosis is the clinical history, for the longer
the time interval between the previous pregnancy and the
curettage material, the more likely is it that the infiltrating
cytotrophoblastic cells are neoplastic in nature. Findings
suggestive of an exaggerated placental site reaction include
the presence of true decidua and placental villi, the pres-
ence of a relatively large number of multinucleated tropho-
blastic cells, and an absence of mitotic figures. By contrast,
features indicating a diagnosis of placental site trophoblas-
tic tumor are the presence of sheets or confluent masses of
cytotrophoblastic cells, a paucity of multinucleated cells,
and the presence of mitotic figures.
The distinction between a placental site tumor and a
choriocarcinoma in curettage material can also prove diffi-
cult, but a preponderance of cytotrophoblast, and a relative
or absolute absence of syncytiotrophoblast, and a haphaz-
ard juxtaposition of cytotrophoblast and any multinucle-
ated cells that may be present all point to a diagnosis of
placental site trophoblastic tumor. In equivocal cases, stain-
ing for hPL and hCG may help to resolve the dilemma, the
true choriocarcinoma showing extensive positive staining
for hCG and the placental site tumor tending to stain pre-
dominantly for hPL and only focally for hCG.
Occasionally, a placental site tumor in curettage material
may be confused with a poorly differentiated endometrial
adenocarcinoma, a squamous cell carcinoma of the cervix, a
clear cell carcinoma, or an epithelioid leiomyosarcoma. The
positive staining of a placental site tumor for cytokeratins,
hPL, and hCG, together with negative staining for vimentin
and actin, will almost invariably resolve such diagnostic
difficulties.
Assessment of the degree of malignancy of a placental
site trophoblastic tumor is difficult, largely because criteria
for distinguishing benign from malignant placental site tro-
phoblastic tumors have not yet been firmly established. It
has been claimed that tumors with a mitotic count of less
than two per ten high-power fields behave in a benign fash-
ion,43 but, nevertheless, occasional neoplasms with only two
mitotic figures per ten high-power fields have pursued a
highly malignant course and proved fatal.45–47 There is little
doubt that a placental site trophoblastic tumor with a high
mitotic count should be considered as being at least poten-
tially malignant, but a false sense of security should not
be engendered by a low mitotic count. Whether or not the
degree of atypia in the neoplasm is of prognostic impor-
tance is currently undecided.
Placental site tumors have been considered as resistant
to chemotherapy, but some success is now being achieved
with intensive regimens.
Epithelioid trophoblastic tumor
The epithelioid trophoblastic tumor is a recently introduced
entity48 which has not yet been fully delineated. The clinical
features and gross appearances of this neoplasm are very
similar to those of a placental site trophoblastic tumor.
Histologically, mononuclear trophoblastic cells form nests,

cords, and solid masses; the tumors grow in a nodular
expansile fashion and lack the infiltrative growth pattern
of placental site trophoblastic tumor. There is often a sur-
rounding lymphocytic infiltrate, and the cords and nests are
intimately associated with eosinophilic fibrillary hyaline-
like material and necrotic debris. Mitotic figures are usually
sparse. The cells stain positively for epithelial markers,
inhibin-alpha and, in a patchy focal fashion, for hPL, hCG,
and Mel-CAM.
This tumor is distinguished from a placental site tropho-
blastic tumor by its non-infiltrative nodular growth pattern,
the presence of hyaline and necrotic debris, and the patchy,
as opposed to diffuse, staining for hCG, hPL, and Mel-
CAM. Distinction from a squamous cell carcinoma, often
difficult on purely histological grounds, rests upon the posi-
tive staining for trophoblastic markers. The clinical course
of the epithelioid trophoblastic tumor is not yet fully de-
fined, but appears to be very similar to that of the placental
site trophoblastic tumor.
References
1. Szulman AE, Surti U (1978) The syndromes of hydatidiform mole.
I. Cytogenetic and morphologic correlations. Am J Obstet Gynecol
131:665–671
2. Szulman AE, Surti U (1978) The syndromes of hydatidiform mole.
II. Morphologic evolution of the complete and partial mole. Am J
Obstet Gynecol 132:20–27
3. Paradinas FJ, Fisher RE, Browne P, Newland ES (1997) Diploid
hydatidiform moles with fetal red blood cells in molar villi. 1.
Pathology, incidence and prognosis. J Pathol 181:183–188
4. Paradinas FJ (1994) The histological diagnosis of hydatidiform
moles. Curr Diag Pathol 1:24–31
5. Paradinas FJ, Browne P, Fisher RA, et al. (1996) A clinical, histo-
logical and flow cytometric study of 149 complete moles, 146 par-
tial moles and 107 non-molar hydropic abortions. Histopathology
28:101–109
6. Bewtra C, Frankforter S, Marcus JN (1997) Clinicopathologic dif-
ferences between diploid and tetraploid complete hydatidiform
moles. Int J Gynecol Pathol 16:239–244
7. Steller MA, Genest DR, Bernstein MR, et al. (1994) Clinical fea-
tures of multiple conception with partial or complete molar preg-
nancy and coexisting fetuses. J Reprod Med 39:147–154
8. Jacobs PA, Szulman AE, Funkmouska J, et al. (1982) Human
triploidy: relationship between paternal origin of the additional
haploid complement and development of partial hydatidiform
mole. Ann Hum Genet 46:223–231
9. Vejerslev LO, Fisher RA, Surti U, Walke N (1987) Hydatidiform
mole: cytogenetically unusual cases and their implications for the
present classification. Am J Obstet Gynecol 157:180–184
10. Lage JM, Popek EJ (1993) The role of DNA flow cytometry in
evaluation of partial and complete hydatidiform moles and hydro-
pic abortions. Semin Diagn Pathol 10:267–274
11. Sunde L, Vejerslev LO, Jensen MP, et al. (1994) Genetic analysis
of repeated, biparental, diploid, hydatidiform moles. Cancer Genet
Cytogenet 66:16–22
12. Barton SC, Surani MAH, Norris ML (1984) Role of paternal and
maternal genomes in mouse development. Nature 311:374–376
13. Ariel I, Lustig O, Oyer CE, et al. (1994) Relaxing of imprinting in
trophoblastic disease. Gynecol Oncol 53:212–219
14. Lage JM, Bagg A (1996) Hydatidiform moles: DNA flow
cytometry, image analysis and selected topics in molecular biology.
Histopathology 28:379–382
15. Conran RM, Hitchcock GL, Popek EJ, et al. (1993) Diagnostic
considerations in molar gestations. Hum Pathol 24:41–48
16. Howat AJ, Beck S, Fox H, et al. (1993) Can histopathologists
reliably diagnose molar pregnancy? J Clin Pathol 46:599–602
77
17. Lage JM, Mark SD, Roberts DJ, et al. (1992) A flow cytometric
study of 137 fresh hydropic placentas: correlation between types of
hydatidiform moles and nuclear DNA ploidy. Obstet Gynecol
79:403–410
18. van de Kaa CA, Robben JCM, Hopman AHN, et al. (1995) Com-
plete hydatidiform mole in twin pregnancy: differentiation from
partial mole with interphase cytogenetic and DNA cytometric
analysis on paraffin embedded tissues. Histopathology 26:123–
129
19. Genest DG, Laborde O, Berkowitz RS, et al. (1991) A clinical-
pathologic study of 153 cases of complete hydatidiform mole
(1980–1990): histologic grade lacks prognostic significance. Obstet
Gynecol 77:111–115
20. Cheung AN, Ngan HY, Collins RJ, Wong YL (1994) Assessment
of cell proliferation in hydatidiform mole using monoclonal anti-
body MIB1 to Ki-67 antigen. J Clin Pathol 47:601–604
21. Cameron B, Gown AM, Tamimi HK (1994) Expression of c-erb-B-
2 oncogene product in persistent gestational trophoblastic disease.
Am J Obstet Gynecol 170:1616–1621
22. Hemming JD, Quirke P, Womack C, et al. (1988) Flow cytometry
in persistent trophoblastic disease. Placenta 9:615–621
23. Wake N, Fujino T, Hoshi S, et al. (1987) The propensity to malig-
nancy of dispermic heterozygous moles. Placenta 8:318–326
24. Mutter GL, Pomponio RJ, Berkowitz RS, Genest DR (1993) Sex
chromosome composition of complete hydatidiform mole: rela-
tionship to metastasis. Am J Obstet Gynecol 168:1547–1551
25. Cheung AN, Sit AS, Chung LP, et al. (1994) Detection of heterozy-
gous XY complete hydatidiform mole by chromosome in situ hy-
bridization. Gynecol Oncol 55:386–392
26. Mueller UW, Hawes CS, Wright AE, et al. (1990) Isolation of fetal
trophoblast cells from peripheral blood of pregnant women.
Lancet 336:197–200
27. Attwood HD, Park WW (1961) Embolism to the lungs by tropho-
blast. Br J Obstet Gynaecol 68:611–617
28. Bagshawe KD (1973) Recent observations related to chemo-
therapy and immunology of gestational choriocarcinoma. Adv
Cancer Res 18:231–263
29. Buckley JD, Henderson BE, Morrow CP, et al. (1988) Case-
control study of gestational choriocarcinoma. Cancer Res 48:1004–
1010
30. Chaganti RSK, Koduru PRK, Chakraborty R, Jones WB (1990)
Genetic origin of a trophoblastic choriocarcinoma. Cancer Res
50:6330–6333
31. Azuma C, Saji F, Nobunaga T, et al. (1990) Studies of the patho-
genesis of choriocarcinoma by analysis of restriction fragment
length polymorphisms. Cancer Res 50:488–491
32. Osada H, Dawata M, Yamada M, et al. (1991) Genetic identifica-
tion of pregnancies responsible for choriocarcinoma after multiple
pregnancies by restriction fragment length polymorphism analysis.
Am J Obstet Gynecol 154:682–688
33. Suzuki T, Goto S, Nawa A, et al. (1993) Identification of the
pregnancy responsible for gestational trophoblastic disease by
DNA analysis. Obstet Gynecol 82:629–634
34. Fisher RA, Soteriou BA, Meredith FJ, et al. (1995) Previous hyda-
tidiform mole identified as the causative pregnancy of choriocarci-
noma following birth of normal twins. Int J Gynecol Cancer
5:54–70
35. Arima T, Imamura T, Amada S, et al. (1995) Genetic origin
of malignant trophoblastic neoplasms. Cancer Genet Cytogenet
73:95–102
36. Wake N, Matsuda T, Arima T, et al. (1998) Genetics of gestational
trophoblastic disease. CME J Gynecol Oncol 3:5–13
37. Fisher RA, Newlands ES, Jeffreys AJ, et al. (1992) Gestational and
non-gestational trophoblastic tumors distinguished by DNA analy-
sis. Cancer 69:839–845
38. Driscoll SG (1963) Choriocarcinoma: an “incidental finding”
within a term placenta. Obstet Gynecol 21:96–102
39. Fox H, Laurini RN (1998) Intraplacental choriocarcinoma: a re-
port of two cases. J Clin Pathol 41:1085–1088
40. Hallam LA, Mclaren KM, El-Jabbour JN, et al. (1990)
Intraplacental choriocarcinoma: a case report. Placenta 11:247–
251
41. Heifetz SA, Csaja J (1992) In situ choriocarcinoma arising in par-
tial hydatidiform mole: implications for risk of persistent tropho-
blastic disease. Pediatr Pathol 12:601–611
78
42. Acosta-Sison H (1955) Can the implanting trophoblast of the fertil-
ized ovum develop immediately into choriocarcinoma? Am J
Obstet Gynecol 69:442–444
43. Young RH, Kurman RJ, Scully RE (1988) Proliferations and
tumors of the placental site. Semin Diagn Pathol 5:223–237
44. Baergen RN, Rutgers JL (1997) Trophobblastic lesions of the
placental site. Gen Diagn Pathol 143:143–158
45. Echstein RP, Paradinas FJ, Bagshawe KD (1982) Placental
site trophoblastic tumour (trophoblastic pseudotumour): a study of
four cases requiring hysterectomy, including one fatal case. Histo-
pathology 6:221–226
46. Fukunaga M, Ushigome S (1993) Metastasizing placental tropho-
blastic tumor: an immunohistochemcal and flow cytometric study
of two cases. Am J Surg Pathol 17:1003–1010
47. How J, Scurry J, Grant P, et al. (1995) Placental site trophoblastic
tumor: report of three cases and review of the literature. Int J
Gynecol Cancer 5:154–249
48. Shih I-M, Kurman RJ (1998) Epithelioid trophoblastic tumor: a
neoplasm distinct from choriocarcinoma and placental site tropho-
blastic tumor simulating carcinoma. Am J Surg Pathol 22:1393–
1403