Seminar

Gestational trophoblastic disease

Michael J Seckl, Neil J Sebire, Ross S Berkowitz

Gestational trophoblastic disease encompasses a range of pregnancy-related disorders, consisting of the premalignant disorders of complete and partial hydatidiform mole, and the malignant disorders of invasive mole, choriocarcinoma, and the rare placental-site trophoblastic tumour. These malignant forms are termed gestational trophoblastic tumours or neoplasia. Improvements in management and follow-up protocols mean that overall cure rates can exceed 98% with fertility retention, whereas most women would have died from malignant disease 60 years ago. This success can be explained by the development of eective treatments, the use of human chorionic gonadotropin as a biomarker, and centralisation of care. We summarise strategies for management of gestational trophoblastic disease and address some of the controversies and future research directions.

Lancet 2010; 376: 71729 Published Online July 28, 2010 DOI:10.1016/S01406736(10)60280-2 Department of Cancer Medicine (Prof M J Seckl FRCP) and Department of Histopathology (N J Sebire FRCPath), Charing Cross Gestational Trophoblastic Disease Centre, Charing Cross Hospital Campus of Imperial College London, London, UK; and New England Trophoblastic Disease Center, Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, Harvard Medical School, Brigham and Womens Hospital, Dana-Farber Cancer Institute, Boston, MA, USA (R S Berkowitz MD) Correspondence to: Prof Michael J Seckl, Department of Cancer Medicine, Charing Cross Gestational Trophoblastic Disease Centre, Charing Cross Hospital Campus of Imperial College London, London W6 8RF, UK m.seckl@imperial.ac.uk

Introduction
Hippocrates was probably the rst to describe gestational trophoblastic disease around 400 BC in his description of dropsy of the uterus.1 Although other observations have been made since, Marchand rst associated hydatidiform mole with pregnancy in 1895.1 Healthy trophoblastic tissue aggressively invades the endometrium and develops a rich uterine vasculature, generating an intimate connection between the fetus and the mother known as the placenta. Invasion is one of the distinct features of malignant disease, and healthy trophoblast can be detected by PCR in the maternal circulation.2 Fortunately, malignant-like behaviour is tightly controlled in healthy trophoblast. However, in gestational trophoblastic disease the regulatory mechanisms fail, resulting in tumours that are highly invasive, metastatic, and very vascular. In this Seminar we discuss the epidemiology, origins, pathological changes, and clinical behaviour of the various forms of gestational trophoblastic disease.

gestations, the risk of a third mole is 1520%,11,14,15 and the risk is not decreased by change of partner.16 Some repeat molar pregnancies are due to familial or sporadic biparental molar disease (gure 1). The frequency of choriocarcinoma or placental-site trophoblastic tumour is less well known, since these diseases can arise after any type of pregnancy.17,18 Choriocarcinoma develops in around one in 50 000 deliveries,19 and placental-site trophoblastic tumour accounts for about 02% of cases of gestational trophoblastic disease in the UK.20 The risk of gestational trophoblastic neoplasia might also be linked to hormonal factors, since women with menarche after 12 years of age, light menstrual ow, and previous use of oral contraceptives are at increased risk.21,22 Additionally, risk of malignant disease after hydatidiform mole has been associated with oral contraceptive use (if started when human chorionic gonadotropin [hCG] concentrations are raised) in some23 but not all24 studies.

Causes and genetics Epidemiology

Gestational trophoblastic disease arises more frequently in Asia than in North America or Europe,3,4 which could be due to dierences in prevalence, discrepancies between hospital-based and population-based data, or disparity in availability of central pathology review. In the UK, all patients are included on a national register, with central pathology review; and the incidence of complete hydatidiform mole is around one per 1000 pregnancies and three per 1000 for partial hydatidiform mole.5 Other developed countries report similar data.6 The incidence of molar pregnancy has decreased in South Korea from 44 cases per 1000 births in the 1960s to 16 cases per 1000 births in the 1990s,7 possibly because of improved socioeconomic conditions and dietary changesespecially since ndings from studies in animals show that diet can reset the genetic imprint.8 Additionally, an increased risk of molar pregnancy is associated with reduced consumption of dietary carotene9,10 and animal fat,9 and advanced maternal age.3,1113 Ova from older women are more susceptible to abnormal fertilisations than are those from younger women. After a molar pregnancy, the risk of further complete and partial mole rises to 12%.11,14,15 After two molar
www.thelancet.com Vol 376 August 28, 2010

In most cases, complete hydatidiform mole usually arises when an ovum without maternal chromosomes

Search strategy and selection criteria We searched the Cochrane Library, Medline (via PubMed, Internet Grateful Med, OVID, and Knowledgender), for meta-analyses, previous systematic reviews, cohort studies (and when appropriate comparison groups), and case-control studies published in English between 1980 and 2010, with the keywords trophoblastic disease, GTD, GTN, choriocarcinoma, molar pregnancy, hydatidiform mole, placental site trophoblastic tumor, genetics, epidemiology, pathology, treatment, chemotherapy, methotrexate, actinomycin D, dactinomycin, cisplatin, paclitaxel, high-dose, management, risk factors, hCG, imaging, ultrasound, PET, CT, MRI, prognosis, and staging. We included results presented at the 15th International Society for the Study of Trophoblastic Diseases meeting in November, 2009, in Cochin, India. Reference lists from all previous publications were scanned to nd any publications not already identied by our electronic search strategy.

717

Seminar

is fertilised by one sperm that then duplicates its DNA, resulting in a 46XX androgenetic karyotype, in which all chromosomes are paternally derived.2527 About 10% of complete moles are 46XY,28 arising from fertilisation by two sperm (gure 1). Although nuclear DNA is entirely paternal, mitochondrial DNA remains maternal in origin.29 Findings from some studies30 show that patients

Androgenetic diploid (monospermic)

X

CHM

XX

Y X

Androgenetic diploid (dispermic)

CHM

XY

with recurrent disease can have biparental molar rather than typical androgenetic disease, which might be familial or sporadic. Genetic studies in such families showed that the related genes are at chromosome 19q13.313.4,31 and subsequent analysis noted NLRP7 mutations in this region.32 The function of the normal protein and the mechanism by which mutations are associated with imprinting abnormalities and gestational trophoblastic disease are unknown.33 Data show clustering of mutations in the leucine-rich region of NLRP7 (gure 2), suggesting that this region is crucial for normal function.34 Some androgenetic diploid complete moles and possibly even triploid partial hydatidiform moles might also carry NLRP7 mutations,35 but conrmation from large studies is needed. Partial hydatidiform moles are almost always triploid (gure 1), and they result from fertilisation of a seemingly healthy ovum by two sperm;3638 diploid partial moles probably do not exist, with most reported cases being misdiagnosed complete moles.39

Pathology
X X

Biparental triploid

PHM

XXX

biCHM

XX

Chromosome 19q: NLRP7 (NALP7) defect Sporadic or hereditary Autosomal recessive

Figure 1: Karyotype derivation of complete and partial hydatidiform moles and rare biparental repetitive complete hydatidiform mole CHM=complete hydatidiform mole. PHM=partial hydatidiform mole. biCHM=rare biparental complete hydatidiform mole. Paternal (black) and maternal (red) derived genes are shown.
R693Q R693W R693P P716A R721W L398R P651S C761Y N913S

E113GfsX7 T61TfsX7 PYD Y318CfsX7 NACHT

R432X

E570X

P716LfsX21 L6776PfsX6 Leucine-rich region

Figure 2: Domain structure of NLRP7 and identied mutations noted in 17 families with biparental repetitive hydaditiform moles Predicted protein domains include PYRIN (PYD), NACHT, and a leucine-rich region. The nine missense aminoacid substitutions or mutations are shown above the protein and seven nonsense mutations are shown below the protein. There seems to be some clustering of mutations in the leucine-rich region.

All gestational trophoblastic disease is derived from the placenta. Hydatidiform moles and choriocarcinoma arise from villous trophoblast and placental-site trophoblastic tumours from interstitial trophoblast. Most complete and partial hydatidiform moles have distinctive morphological characteristics, although diagnostic criteria have changed because evacuation is done earlier in gestation (median 89 weeks in the UK). First-trimester complete moles show a characteristic abnormal budding villous structure with trophoblast hyperplasia, stromal karyorrhectic debris, and collapsed villous blood vessels. By contrast, early partial moles show patchy villous hydrops with scattered abnormally shaped irregular villi, trophoblastic pseudoinclusions, and patchy trophoblast hyperplasia (gure 3).4042 Morphological distinction of non-molar miscarriage from partial hydatidiform mole can be dicult, since villous dysmorphism can be present but without the characteristic trophoblast hyperplasia that is noted in partial mole. Ancillary techniques are needed in some cases to dierentiate non-molar miscarriage from hydatidiform mole, including immunostaining for P57kip2, the product of CDKN1C. P57kip2 is expressed by the maternal allele and is visible on histology as nuclear staining of cytotrophoblast and villous mesenchyme in placenta of all gestations apart from androgenetic complete mole.43,44 Additionally, ploidy analysis by in-situ hybridisation or ow cytometry can distinguish diploid from triploid conceptions, helping to diagnose partial mole, but is unable to distinguish complete mole from diploid non-molar miscarriage, or molar versus non-molar triploidy, which necessitate molecular investigations.18,4547 Choriocarcinomas are malignant hCG-producing epithelial tumours with central necrosis and a characteristic biphasic architecture recapitulating cytotrophoblast-like
www.thelancet.com Vol 376 August 28, 2010

718

Seminar

1 mm

05 mm

01 mm

01 mm

Figure 3: Gestational trophoblastic disease (A) Complete hydatidiform mole (magnication 20). (B) Partial hydatidiform mole (magnication 40). (C) Placental-site trophoblastic tumour (magnication 200). (D) Choriocarcinoma (magnication 100). All forms have abnormal trophoblast proliferation, which is associated with dysmorphic chorionic villi in complete and partial hydatidiform moles, but villi and abnormal trophoblast invasion are not noted in placental-site trophoblastic tumour or choriocarcinoma.

cells and multinucleate, pleomorphic syncytiotrophoblastlike regions; however, mononuclear cells predominate in some cases, especially after chemotherapy.48 Intraplacental choriocarcinoma does occur and is probably the source of metastatic disease after term pregnancies. Most cases of neonatal choriocarcinoma result from metastatic spread from an intraplacental choriocarcinoma.49 Placental-site trophoblastic tumours are the malignant equivalent of extravillous, interstitial implantation site-like trophoblast and form uterine lesions with less haemorrhage and necrosis, and lower hCG concentrations than does choriocarcinoma.48,50 Epithelioid trophoblastic tumoura variant of placental-site trophoblastic tumour with similar clinical behaviour but distinctive hyalinisationhas been reported, but data for this disease are sparse.51 Accurate measurement of hCG is key to eective management of gestational trophoblastic disease, several cancers, and pregnancy. Panel 1 summarises some of the issues related to hCG measurements.

Diagnostic ultrasound for molar pregnancy

Patients with complete hydatidiform mole most commonly present with vaginal bleeding in early
www.thelancet.com Vol 376 August 28, 2010

pregnancy. Previously reported features such as anaemia, uterine enlargement, pre-eclampsia, hyperemesis, hyperthyroidism, and respiratory distress are now rare,60 probably because routine use of ultrasonography leads to diagnosis in the rst rather than late-second trimester. Partial hydatidiform moles grow more slowly and present slightly later in the rst or early second trimester than do complete moles, but also manifest with vaginal bleeding or missed or incomplete miscarriages.61,62 Characteristic ultrasonographic scans of complete mole show a uterine cavity lled with a heterogeneous mass (so-called snowstorm), without associated fetal development and with theca lutein ovarian cysts.63,64 However, these features are not visible in the rst trimester and, although some investigators6570 have suggested that ultrasound can be diagnostic of complete mole in early pregnancy, large studies71,72 have shown that only 4060% of cases are detected as molar by sonography in routine clinical practice. Moreover, 10% of cases that are thought to be molar on sonography were diagnosed as nonmolar hydropic abortions on histological review.71 Findings from other reports73 are similar and emphasise
719

Seminar

Panel 1: hCG measurement in gestational trophoblastic disease and cancer hCG comprises an subunit (shared with other members of the glycoprotein hormones, including thyroid-stimulating hormone and luteinising hormone) and a subunit that confers specicity. Consequently, assays designed to specically detect hCG have to target the subunit. In healthy pregnancy, hCG is intact and is hyperglycosylated during the rst trimester. However, in cancer, many other subtypes of -hCG can exist, including free-hCG, -core, nicked free-, or c-terminal peptide.52,53 Therefore, hCG assays used in cancer need to detect all forms of hCG and not only those found in healthy pregnancy. Moreover, the dierent -hCG forms should be detected equally well. Unfortunately, most commercial assays fail to or variably detect all forms of -hCG and therefore are prone to false-negative results in patients with cancer.52,54 Additionally, every assay is susceptible to false-positive results, usually caused by cross-reacting heterophile antibodies. These antibodies do not pass into the urine, so if hCG is present, a false-positive serum value can be excluded in most cases, although other approaches to this problem might be necessary.55 However, the false-negative issue cannot be easily resolved and is potentially clinically important, since a false-negative nding could result in delayed diagnosis or premature withdrawal of chemotherapy, or both.52,54 No commercial assay is licensed for use in cancer diagnosis or management at present, although many are used for this purpose. In the UK, we use a non-commercial rabbit polyclonal antibody that detects all forms of hCG for monitoring of patients with gestational trophoblastic disease. By comparison with other assays, our assay seems to have a low false-negative rate and, since we routinely measure hCG in serum and urine, false-positive results are rare. This assay is not available worldwide and, to our knowledge, the only commercial assay that seems comparably safe to use in gestational trophoblastic disease is the Siemens Immulite (Deereld, IL, USA).53,54 Therefore, a new generation of cancer-specic hCG assays is urgently needed. Findings from recent studies56,57 that used a hyperglycosylated hCG assay suggest that a high ratio of this variant to total hCG can detect malignant forms of gestational trophoblastic disease. These ndings correlate with some biological evidence56 suggesting that hyperglycosylated hCG induces an invasive phenotype in trophoblast and choriocarcinoma cells. Additionally, some preliminary studies58,59 suggest that free--hCG is a marker for placental-site trophoblastic tumour and seminomatous germ cell tumours. Prospective studies in large patient cohorts are needed to dene the role of hyperglycosylated hCG and free--hCG in the management of gestational cancers.
hCG=human chorionic gonadotropin.

Surgical evacuation
Suction curettage is the preferred method of evacuation irrespective of uterine size in patients with suspected hydatidiform mole who want to preserve fertility.78,79 Intraoperative ultrasonography can reduce the risk of uterine perforation. Patients who are rhesus-negative should receive rhesus immunoglobulin at the time of evacuation because rhesus D factor is expressed on trophoblast. Women who are nulliparous should not be given prostanoids to ripen the cervix since these drugs can induce uterine contractions and might increase the risk of trophoblastic embolisation to the pulmonary vasculature.80 Hysterectomy is rarely recommended but might be considered for women who do not want further children or have life-threatening haemorrhage.81,82 Patients should be counselled that, although hysterectomy stops the risk of local invasion, it does not eliminate the possible need for chemotherapy, and monitoring of hCG concentrations should still be done.11 A healthy co-twin can develop alongside a complete or partial hydatidiform mole in one per 20 000100 000 pregnancies (gure 4). Some investigators have suggested that such pregnancies should be terminated because of the low probability of successful outcome and the increased risk of development of malignant disease.83 However, evidence from a case series84 of 77 pregnancies suggests that about 40% of women will deliver a healthy baby without a signicant increase in the risk of malignant transformation of the complete hydatidiform mole. Findings from a study76 of 2800 singleton molar pregnancies lent support to the notion that late evacuation of complete hydatidiform mole is not associated with an increased rate of malignant disease.

Registration for hCG surveillance

All patients with hydatidiform mole should be registered with a specialist centre for hCG surveillance, preferably one that is coordinated nationally. The UK established the rst national service for gestational trophoblastic disease through a combined agreement of the Royal College of Obstetricians and Gynaecologists and the Department of Healths National Commissioning Group. Since 1973, all women with hydatidiform mole or other forms of gestational trophoblastic disease have been registered with one of three centres for hCG monitoring and, if subsequent treatment is necessary, are treated at the Sheeld Trophoblastic Disease Centre (Sheeld, UK) or Charing Cross Hospital Trophoblast Disease Centre (London, UK). Onset of malignant change, termed persistent gestational trophoblastic disease or post-mole neoplasia, is signied by a plateaued or rising hCG concentration. Studies in the UK show that malignant change arises after 15% of complete and 0510% of partial hydatidiform moles.11,18,85 Rates are probably higher in other countries than the UK,86 possibly because of
www.thelancet.com Vol 376 August 28, 2010

that false-positive and false-negative rates are high with ultrasonographyespecially for partial hydatidiform moleand histological examination is necessary to achieve a correct diagnosis.40,74 All products of conception from non-viable pregnancies should undergo histological examination irrespective of ultrasonographic ndings.75 Some women who miscarry or have medical terminations will have had unsuspected molar pregnancies. Because ultrasonography cannot reliably conrm molar disease and histological examination of products might not be done after pregnancy termination, delayed diagnosis of gestational trophoblastic neoplasia, and hence substantial morbidity with the need for complex chemotherapy and surgery, can result.76 Histological examination after every termination would be impracticable, so checking of hCG concentrations at 34 weeks after treatment to ensure a return to a value within the normal range is recommended.77
720

Seminar

dierences in hCG criteria and overdiagnosis of neoplasia, scarcity of whole-population demographics, or a dierence in disease biology, although this explanation is unlikely. Precise surveillance protocols vary by country but principles are alike. In the UK, serum and urine hCG concentrations are measured every 2 weeks until the values are within the normal range, and then urine hCG concentrations are recorded monthly. Patients with normal hCG values within 56 days of uterine evacuation have a reduced risk of development of malignant disease,11,87 and are monitored for 6 months from evacuation date. When the rst hCG reading within the normal range is noted after 56 days, monthly monitoring continues for 6 months.87 Some investigators suggest shortened hCG surveillance, with discontinuation after the rst value within the normal range is measured, especially for women with partial mole in whom the risk of neoplasia is reduced.88,89 Shortened surveillance could enable women to attempt a subsequent pregnancy sooner, but could result in late development of neoplasia with increased morbidity and mortality. Data reported at the 2009 International Society for the Study of Trophoblastic Diseases (ISSTD) world congress from 22 000 women with complete and partial hydatidiform moles in the UK suggest that either form can occasionally develop post-mole neoplasia after the hCG has returned to normal; however, the risk of missed gestational trophoblastic neoplasia can be reduced from one in 800 women to one in 1400 by following present UK guidelines.87 Consequently, UK practice seems good for young women, but older nulliparous women should be made aware of the relative risks and benets of an early pregnancy. During hCG follow-up, patients are encouraged to use reliable contraception, including a combination of methods, although data are conicting about whether oral contraceptives increase the risk for gestational trophoblastic neoplasia. In the UK, practitioners avoid recommending oral contraceptives until hCG is in the normal range.23,24 After completion of hCG monitoring, serum or urine hCG concentrations should be checked 6 weeks and 10 weeks after every pregnancy to ensure no reactivation of previous molar disease.90,91

Complete hydatidiform mole

Normal placenta

Figure 4: Twin pregnancy of a complete hydatidiform mole and healthy co-twin (A) Ultrasonography and (B) MRI.

Panel 2: Indications for chemotherapy for gestational trophoblastic disease in the UK Plateaued or rising hCG concentration after evacuation* Heavy vaginal bleeding or evidence of gastrointestinal or intraperitoneal haemorrhage Histological evidence of choriocarcinoma Evidence of metastases in brain, liver, or gastrointestinal tract, or radiological opacities larger than 2 cm on chest radiograph Serum hCG concentration of 20 000 IU/L or more, 4 weeks or more after evacuation, because of the risk of uterine perforation Raised hCG concentration 6 months after evacuation, even when still decreasing
hCG=human chorionic gonadotropin. *A plateaued hCG concentration is dened as four or more equivalent values of hCG for at least 3 weeks (days 1, 7, 14, and 21), and rising as two consecutive increases in hCG concentration of 10% or more for at least 2 weeks (days 1, 7, and 14).

Treatment
Indications
Panel 2 shows the UK indications for treatment of gestational trophoblastic disease with chemotherapy. These recommendations are similar to those suggested by the International Federation of Gynecology and Obstetrics (FIGO)92 and include a plateaued or rising hCG (the most common reason for treatment), a persistently raised hCG at 6 months after evacuation, or histological diagnosis of choriocarcinoma. However, our experience suggests that the disease is unlikely to remit spontaneously when: the hCG concentration is more than 20 000 IU/L, 1 month after evacuation (also
www.thelancet.com Vol 376 August 28, 2010

associated with an increased risk of uterine perforation); or there are lung or vaginal metastases more than 2 cm in diameter (small lesions can spontaneously regress), or spread to other organs.42,92 Additionally, in the UK, chemotherapy is started to help to stop heavy bleeding that necessitates transfusion, even when the hCG concentration is falling.

Staging and stratication

Most patients who develop gestational trophoblastic neoplasia after hydatidiform mole are detected early by hCG monitoring so detailed investigation is rarely needed. Information on which to base therapy decisions can be obtained from clinical histories, examination, and measurement of serum hCG. Additionally, patients should have doppler pelvic ultrasonography to conrm absence of pregnancy, and to measure the uterine size and volume, spread of disease within the pelvis, and disease vascularity (gure 5). Disease vascularity can
721

Seminar

Prechemotherapy

Postchemotherapy

systems (table 1). Since 2002, all physicians treating gestational trophoblastic neoplasia should use this system to allow comparison of data. The combined prognostic score predicts potential for development of resistance to monochemotherapy with methotrexate or dactinomycin. A score of 06 suggests low risk of resistance and 7 or more indicates high risk. Such disease has almost no chance of being cured with monochemotherapy and needs multidrug treatment. Anatomical staging does not aid therapeutic choices but helps clinicians to compare results between centres.

Prechemotherapy

Postchemotherapy

Low-risk disease
About 95% of patients with hydatidiform mole who develop neoplasia are at low risk of resistance (score 06). In patients with stage I disease that is seemingly conned to the uterine cavity, the use of second dilatation and curettage to reduce the need for chemotherapy is controversial.98101 Results from the UK suggest that secondary surgery is of no value when hCG concentrations are greater than 5000 IU/L, since more than 50% of such patients will need chemotherapy.101 The low eectiveness of second surgery and small risks of infection, haemorrhage, and uterine perforation should be measured against the almost 100% cure rate and comparative safety of chemotherapy. Some patients with stage I neoplasia who do not want further children request hysterectomy, which can be technically dicult with these highly vascular tumours and does not completely obviate the need for chemotherapy. For most low-risk patients with gestational trophoblastic neoplasia, monochemotherapy with methotrexate or dactinomycin is the preferred treatment. Many regimens have been used, showing a 5090% chance of induction of remission in non-randomised, mostly retrospective, studies.102 The wide variability results from dierences in dose, frequency, route of administration, and the criteria used to select patients for therapy.19,95 Some investigators suggest that intensive therapy given daily for 58 days every 2 weeks is better than treatments given once every 2 weeks;103 others suggest that dactinomycin is more likely to induce remission than is methotrexate.104 The few randomised studies102 that addressed some of these issues were underpowered and compared regimens that are not frequently used internationally. Patients in whom rst-line therapy failsgenerally because of resistance can be easily salvaged with second-line or occasionally third-line chemotherapy, so overall survival is nearly 100%.105 Since survival is so high, patients should be given the least toxic therapy rst to avoid exposure to more harmful treatments. The regimen of methotrexate (50 mg intramuscularly every 48 h for four doses) with calcium folinate (folinic acid) rescue (15 mg orally 30 h after methotrexate) regimen developed at our UK institute is eective and is well tolerated. Courses are repeated every 2 weeks. Unlike dactinomycin, this regimen does not induce hair loss; thus
www.thelancet.com Vol 376 August 28, 2010

Prechemotherapy

Postchemotherapy

Figure 5: Prechemotherapy and postchemotherapy imaging (A) Doppler ultrasound examination of a complete hydatidiform mole with a vascular uterine mass. (B) Chest radiograph of a patient 9 months after a term delivery with multiple pulmonary metastases. (C) Brain MRI with three obvious haemorrhagic metastases in a patient with seizures 6 months after a term delivery.

suggest patients who are at risk of treatment resistance.93,94 Pulmonary metastases are most common, so chest radiography is essential.95 Chest CT is not needed when ndings from chest radiography is normal, since discovery of micrometastases, which can be seen in about 40% of patients, does not aect outcome.96,97 However, if lesions are noted on chest radiograph, brain MRI and body CT are recommended to exclude more widespread disease aecting, for example, the brain or liver, which would substantially change management. FIGO report data for gestational trophoblastic neoplasia by use of prognostic scoring and anatomical staging
722

Seminar

0 Age (years) Antecedent pregnancy Interval from antecedent pregnancy to chemotherapy (months) hCG concentration (IU/L) Number of metastases Site of metastases Largest tumour mass diameter (cm) Previous chemotherapy <40 Mole <4 <10 0 Lung

1 40 Abortion 46 10<10 14 Spleen, kidney 35

2 Term 712 1010 58 Gastrointestinal tract >5 Monotherapy

4 >12 >10 >8 Brain, liver Combined therapy

hCG=human chorionic gonadotropin. *Patients score is total of individual scores for the eight prognostic factors. Low risk of resistance to monochemotherapy 06, high risk 7 or more. Placental-site trophoblastic tumour should not be scored but needs to be staged. Stage I disease is conned to the uterus; stage II disease extends to the genital tract; stage III disease is spread to lungs with or without extension to the genital tract; and stage IV occupies all other metastatic sites including liver, kidney, spleen, and brain.

Table 1: International Federation of Gynecology and Obstetrics (2000) scoring system for gestational trophoblastic neoplasia, by prognostic factor*

it has been widely used.105107 Most patients are treated at home after a short stay in hospital to monitor any bleeding. About 2% of women have mouth ulcers, sore eyes, or rarely pleuritic or peritoneal pains from serositis.105 Patients who develop resistance to methotrexate with folinic acid rescue can be switched to dactinomycin when hCG concentrations are less than or equal to 100 IU/L or multidrug chemotherapy when concentrations are more than 100 IU/L, which will cure nearly all patients.105,106,108 The UK gestational trophoblastic disease service has increased the hCG concentration at which combination chemotherapy is started to more than 300 IU/L, to reduce the number of women being exposed to greater toxicity. Chemotherapy should be continued until hCG is within the normal range and then for a further 6 weeks (gure 6), helping to eliminate any residual tumour cells and reduce the chance of relapse. Only 30% of patients scoring 56 can be cured with lowrisk therapy.105,106,108,109 Therefore, revision of the FIGO scoring system would be helpful for early identication of the 70% of women in this group who develop resistance to methotrexate with folinic acid rescue and who need more intensive therapy. The amount of vascularisation as detected on doppler ultrasonography could help to provide the necessary additional information.93,94 Furthermore, data108 suggest that women in this category with an hCG concentration of more than 400 000 IU/L are unlikely to be cured by methotrexate with folinic acid rescue, so multidrug treatment should be given from the outset.108

High-risk disease
Most high-risk patients with gestational trophoblastic neoplasia present with many metastases months or years after the causative pregnancy of any type. Symptoms and signs vary with disease location. Patients with brain metastases (gure 5) present with seizures, headaches, or hemiparesis,110 whereas those with lung metastasis or disease in the pulmonary vasculature might have a combination of haemoptysis, shortness of breath, or pleuritic chest pain.111 Menstrual irregularity is not universal, so unless clinicians consider gestational
www.thelancet.com Vol 376 August 28, 2010

trophoblastic neoplasia in the dierential of metastatic disease and measure the serum or urine hCG the diagnosis can be missed. If hCG concentrations are raised, the treating physician should consult the nearest gestational trophoblastic disease centre about management. Imaging investigations should include body CT, brain MRI, and pelvic MRI and Doppler ultrasonography (gure 5). If the brain scan is normal, a lumbar puncture to measure the ratio of cerebrospinal uid to serum hCG (normal <1:60) can help to exclude occult CNS disease.110,112 Investigators should avoid taking a biopsy sample of these highly vascular tumours to prevent life-threatening haemorrhage. However, when a lesion is easily accessible and bleeding can be controlled, taking an excision biopsy sample can be helpful. This is especially important when placental-site or non-gestational tumours might exist, since their management diers from that of gestational choriocarcinoma. Fortunately, placental-site trophoblastic tumour has a distinct histological appearance and comparison of microsatellite polymorphisms in the tumour with DNA from the patient and her partner can establish whether the tumour is gestational.91 The phenotypic appearance of the tumour is not always reliable and rarely nongestational carcinomas can look morphologically very similar to gestational choriocarcinomas and, conversely, these carcinomas can occasionally mimic other epithelial tumours.91,113 Chemotherapy is eective for cure of gestational tumours, whereas the chance of survival from a non-gestational tumour depends on the primary site. Patients scoring 7 or more on the FIGO system (table 1) are at high risk of developing drug resistance and are very unlikely to be cured with monochemotherapy. Consequently, several dierent multidrug therapies have been developed, including methotrexate, folinic acid, and dactinomycin (MFA); methotrexate, dactinomycin, cyclophosphamide, doxorubicin, melphalan, hydroxycarbamide, and vincristine (CHAMOCA);114 methotrexate, dactinomycin, and cyclophosphamide (MAC); and etoposide, methotrexate, and dactinomycin (EMA).115 At our UK institute, a regimen was developed consisting of alternating EMA every week with cyclophosphamide and
723

Seminar

100 000 MTXFA Serum hCG concentration (IU/L) 10 000 Uterine evacuation of CHM MTXFA MTXFA 100 6 weeks (3 treatment cycles) with a normal hCG concentration MTXFA MTXFA Stop RX 4

1000

10

0 1 2 Time (months) 3

Figure 6: Response to low-risk chemotherapy measured by hCG tumour marker concentration After uterine evacuation of a complete hydatidiform mole, hCG remained plateaued, suggesting persistent gestational trophoblastic disease or neoplasia, so the patient started treatment with methotrexate and folinic acid. Time and type of intervention are shown above the line. Therapy was continued for 6 weeks after normal hCG concentration (<5 IU/L) was recorded. hCG=human chorionic gonadotropin. CHM=complete hydatidiform mole. MTXFA=methotrexate with folic acid rescue. STOP RX=stop treatment.

Schedule Day 1 (EMA) Etoposide Dactinomycin Methotrexate Day 2 (EMA)* Etoposide Dactinomycin Day 8 (CO) Vincristine 1 mg/m intravenous bolus (maximum 2 mg) Cyclophosphamide 600 mg/m intravenous infusion for 30 min
EMA=etoposide, methotrexate, and dactinomycin. CO=cyclophosphamide and vincristine. EMA alternates with CO every week. *Reduction of the EMA by omission of the day 2 dose of etoposide and dactinomycin is only exceptionally needed to avoid extended intervals between courses caused by myelosuppression. This problem is usually overcome with lgrastim and increased frequency (four doses daily) and duration (up to 4 days) of folinic acid rescue. Rescue begun 24 h after start of methotrexate infusion.

100 mg/m by intravenous infusion for 30 min 05 mg intravenous bolus 300 mg/m by intravenous infusion for 12 h 100 mg/m by intravenous infusion for 30 min 05 mg intravenous bolus

Folinic acid rescue 15 mg intramuscular or orally every 12 h for four doses

819905) in 272 patients at our UK institute.119 Although these results were good, long-term survival was only 27% when there were metastases in the liver, 70% with brain metastases, and 10% with both sites of metastasis.119,122 Why these patients have adverse outcomes is unclear, but most had not had previous hydatidiform mole, were not registered for follow-up, and therefore presented with widespread disease. Furthermore, many deaths happened soon after admission from haemorrhage or metabolic results of overwhelming disease. Indeed, when deaths within 4 weeks (before adequate chemotherapy can be given) were excluded, survival of patients with brain metastases was equivalent to that for other patients.110 This eect may also apply to liver metastases; of 37 patients with liver metastases treated between 1977 and 2005 at our UK institute, overall survival increased to about 50% at 5 years, but when early deaths were excluded, survival was nearly 70%.123 Besides disease extent, type and duration of antecedent pregnancy and previous chemotherapy are associated with poor outcome.117,124 To reduce early deaths in patients with very advanced disease, we noted that starting chemotherapy with lowdose etoposide (100 mg/m) and cisplatin (20 mg/m) for 2 days, combined with dexamethasone 24 mg in 24 h is helpful to reduce tumour oedema. Further details about management and modications to treatment needed for patients with very advanced disease and dicult clinical situations such as brain metastasis and pulmonary failure have been reviewed elsewhere.110,125 As in low-risk disease, therapy is continued for 6 weeks of normal hCG values in patients with high-risk disease or 8 weeks if poor prognostic features such as liver or brain metastases are present.19 Reimaging is then done to document the post-treatment appearance for future comparison. Removal of residual masses is unnecessary since it does not reduce risk of recurrence, which is less than about 3%.126128

Drug-resistant disease
Patients with gestational trophoblastic neoplasia who progress during or after primary chemotherapy still have excellent outcomes, with about 100% of low-risk and 84% of high-risk patients being cured,128 partly because relapse is detected early by hCG monitoring so disease volume is small. PET scanning with F-uorodeoxyglucose can help to identify the site of active disease to aid surgical resection and cure.129 The half-life for hCG is 48 h or less after surgery if the disease has been completely removed.130 However, when surgery is not possible or hCG concentrations decrease inappropriately, several salvage regimens have been created or adapted from the germ-cell tumour setting.131 At our UK institute we developed a regimen combining etoposide with cisplatin (EP) alternating every week with EMA that omitted the second day of etoposide and dactinomycin.132 Survival is more than 80% but toxic eects are substantial.132 Several patients with drugresistant tumours have responded to gemcitabine133 or
www.thelancet.com Vol 376 August 28, 2010

Table 2: EMA-CO chemotherapy regimen for patients at high risk of monochemotherapy resistance

vincristine (CO; table 2).116 This regimen is widely used worldwide,115 since it seems eective with predictable and easily managed short-term toxic eects. The Korean gestational trophoblastic disease centre noted in a retrospective comparison117 that MFA had a remission rate of 63% (31 of 49 patients), MAC 68% (27 of 40), CHAMOCA 71% (32 of 45), and EMA-CO 91% (87 of 96). The EMA-CO regimen necessitates one overnight stay every 2 weeks, causes reversible alopecia, and is myelosuppressive (although support with lgrastim helps to maintain neutrophil count and treatment intensity, and avoids neutropenic febrile episodes).19 Cumulative 5-year survival of patients given EMA-CO is between 75% and 90%,116121 and was 862% (95% CI
724

Seminar

Schedule Day 1 (TP) Dexamethasone Dexamethasone Cimetidine Chlorphenamine Paclitaxel Mannitol Cisplatin Post-hydration Day 15 (TE) Dexamethasone Dexamethasone Cimetidine Chlorphenamine Paclitaxel Etoposide 20 mg oral (12 h before paclitaxel) 20 mg oral (6 h before paclitaxel) 30 mg in 100 mL normal saline intravenous for 30 min 10 mg intravenous bolus 135 mg/m in 250 mL normal saline intravenous for 3 h 150 mg/m in 1 L normal saline intravenous for 1 h 20 mg orally (12 h before paclitaxel) 20 mg orally (6 h before paclitaxel) 30 mg in 100 mL normal saline intravenous for 30 min 10 mg intravenous bolus 135 mg/m in 250 mL normal saline intravenous for 3 h 10% in 500 mL intravenous for 1 h 60 mg/m in 1 L normal saline intravenous for 3 h 1 L normal saline, 20 mmol potassium chloride, and 1 g magnesium sulphate intravenous for 2 h Year 1 16 weeks 26 months 712 months Year 2 Year 3 Year 4 Year 5 After year 5

Urine

Blood

Every week Every 2 weeks Every 2 weeks Every 4 weeks Every 8 weeks Every 3 months Every 4 months Every 6 months

Every week Every 2 weeks

Table 4: Sampling protocol for hCG concentration in all patients after initial chemotherapy (monotherapy or combination) and following relapse treatment with gestational trophoblastic neoplasia in the UK

TP=paclitaxel-cisplatin. TE=paclitaxel-etoposide.

Table 3: TP-TE schedule for relapsed gestational trophoblastic neoplasia

paclitaxel-based single-agent or combination therapy.134138 An alternating combination of paclitaxel-cisplatin and paclitaxel-etoposide (TP-TE) every 2 weeks (table 3) seems from non-randomised studies to be much better tolerated than is EP-EMA, and is eective in patients with relapsed or refractory neoplasia.138 On the basis of these results the ISSTD has recently proposed a randomised trial of TE-TP versus EP-EMA to assess optimum therapy for patients relapsing after non-cisplatin-based combination therapies such as EMA-CO. High-dose chemotherapy with peripheral stem-cell transplantation does not cure many patients with refractory disease,139 so selection of patients needs to improve.140,141

Placental-site trophoblastic tumour

Placental-site trophoblastic tumours grow more slowly, metastasise later, more commonly involve lymph-nodes, and produce less hCG than do choriocarcinomas.142,143 However, as with choriocarcinoma, this disease can arise after any type of pregnancy, including partial mole17 and most patients present with abnormal vaginal bleeding.20 The disorder should be suspected if hCG concentrations are low for the volume of disease present on imaging and free--hCG values are high, but these features are not diagnostic.58,59 Histological analysis is needed to substantiate the diagnosis. In a population-based series,20 Schmid and colleagues examined prognostic features for placental-site trophoblastic tumours in 62 cases over 30 years, representing 02% of UK gestational trophoblastic disease. By univariate analysis, stage, hCG, mitotic index, and duration of more than 4 years from preceding pregnancy were prognostic, but the FIGO score was not.142144 Only time from previous pregnancy to rst treatment remained predictive of
www.thelancet.com Vol 376 August 28, 2010

survival on multivariate analysis: all 13 patients with an interval of 48 months or more died, and 48 of 49 with an interval of less than 48 months survived. This eect was not explained by dierences in disease stage or hCG concentrations, but might suggest a biological switch in tumours after this time.20 Management of placental-site trophoblastic tumour diers from that for choriocarcinoma. Patients with metastatic disease need combination chemotherapy (eg, EP-EMA) until 8 weeks of normal hCG concentrations are recorded.20 Unlike choriocarcinoma, residual masses are removed surgically, including the uterus, which can contain microscopic disease. The need for surgery might cause diculties for management of stage I disease.145 The safest option is hysterectomy with sampling of pelvic lymph nodes and ovarian conservation, unless the patient has a family history of ovarian cancer or is postmenopausal. In the absence of sucient data for adjuvant therapy, we advocate 8 weeks of EP-EMA or TE-TP when there are poor risk factors such as disease presentation beyond 4 years from the antecedent pregnancy. However, young nulliparous women generally have a strong desire to preserve fertility, especially when there seems to be a focal abnormality in the uterus. Although uterine-sparing surgery is possible,145148 multifocal microscopic uterine disease can happen145 and could compromise survival. Appropriate counselling is needed. Placental-site disease is so rare that the optimum treatment will probably never be identied, but the ISSTD has formed an international database to pool information about cases.

Postchemotherapy sequelae and follow-up

Most patients recover from even intensive chemotherapy within weeks or months, and nearly all side-eects such as alopecia are reversible. Fertility is an important issue for patients with gestational disease. Fortunately, although EMA-CO advances the menopause date by 3 years,149 fertility is not otherwise aected by either methotrexate with folinic acid rescue or EMA-CO chemotherapy. Furthermore, the pregnancy rate is more than 83% after either treatment,150 and the incidence of congenital malformations is not increased.150,151
725

Seminar

However, patients are advised not to become pregnant until 12 months after completion of chemotherapy to reduce the potential teratogenicity and to avoid confusion between a new pregnancy and relapsed disease as the cause of increasing hCG values. Between 1973 and 1997, 230 women at our UK centre became pregnant during the rst year of follow-up, despite the advice to avoid pregnancy. The risk of relapse, fetal morbidity, and maternal death were not increased; more than 70% continued their pregnancies to term, but one patient relapsed with advanced disease.152 Thus, although we advise women to avoid pregnancy for 1 year after completing chemotherapy, those who do become pregnant will probably have a favourable outcome. Any form of contraception is suitable provided there are no other medical contraindications to their use. When a patient does become pregnant, ultrasonography and other appropriate methods should be used to ensure that the pregnancy is healthy. Follow-up therapy is then discontinued but hCG concentrations should be rechecked 6 weeks and 10 weeks after delivery to ensure no recurrence or new disease. In the UK, we follow up patients for life (table 4), since monitoring is easy to do (urine samples only), cheap, and we are uncertain about when stopping is safe. Late sequelae from chemotherapy are very rare. Findings from a study153 of 15 279 patient-years of follow-up showed no signicant increase in incidence of second tumours after methotrexate therapy. By contrast, 26 patients receiving combination chemotherapy for gestational trophoblastic neoplasia developed another cancer when the expected rate was only 1645which is a signicant dierence.

identied for many cases, its function is unknown. The successful UK model of centralised care for rare diseases should be established worldwide to improve outcomes for gestational trophoblastic disease.
Contributors All authors did a detailed review of published work and contributed to the writing, review, and editing of the report. MJS had access to all the data used to write the report and had nal responsibility for submission. All authors saw and approved the nal version. Conicts of interest We declare that we have no conicts of interest. Acknowledgments MJS and NJS thank the Department of Health, National Commissioning Group, and the Cancer Treatment and Research Trust for their continued support. MJS acknowledges support from the Imperial College Experimental Cancer Medicine Centre and Biomedical Research Centre grants. RSB acknowledges the Donald P Goldstein MD Trophoblastic Registry Endowment Fund and the Choriocarcinoma Research Fund. References 1 Ober WB, Fass RO. The early history of choriocarcinoma. Ann NY Acad Sci 1961; 172: 299426. 2 Mueller UW, Hawes CS, Wright AE, et al. Isolation of fetal trophoblast cells from peripheral blood of pregnant women. Lancet 1990; 336: 197200. 3 Bracken MB. Incidence and aetiology of hydatidiform mole: an epidemiological review. Br J Obstet Gynaecol 1987; 94: 112335. 4 Palmer JR. Advances in the epidemiology of gestational trophoblastic disease. J Reprod Med 1994; 39: 15562. 5 Newlands ES, Paradinas FJ, Fisher RA. Recent advances in gestational trophoblastic disease. Hematol Oncol Clin North Am 1998; 13: 22544. 6 Jeers MD, ODwyer P, Curran B, Leader M, Gillan JE. Partial hydatidiform mole: a common but underdiagnosed condition. A 3-year retrospective clinicopathological and DNA ow cytometric analysis. Int J Gynecol Pathol 1993; 12: 31523. 7 Martin BH, Kim JH. Changes in gestational trophoblastic tumors over four decades. A Korean experience. J Reprod Med 1998; 43: 6068. 8 Waterland RA, Jirtle RL. Early nutrition, epigenetic changes at transposons and imprinted genes, and enhanced susceptibility to adult chronic diseases. Nutrition 2004; 20: 6368. 9 Berkowitz RS, Cramer DW, Bernstein MR, Cassells S, Driscoll SG, Goldstein DP. Risk factors for complete molar pregnancy from a case-control study. Am J Obstet Gynecol 1985; 152: 101620. 10 Parazzini F, La Vecchia C, Mangili G, et al. Dietary factors and risk of trophoblastic disease. Am J Obstet Gynecol 1988; 158: 9399. 11 Bagshawe KD, Dent J, Webb J. Hydatidiform mole in England and Wales 197383. Lancet 1986; 328: 67367. 12 Parazzini F, Mangili G, La Vecchia C, Negri E, Bocciolone L, Fasoli M. Risk factors for gestational trophoblastic disease: a separate analysis of complete and partial hydatidiform moles. Obstet Gynecol 1991; 78: 103945. 13 Sebire NJ, Foskett M, Fisher RA, Rees H, Seckl M, Newlands E. Risk of partial and complete hydatidiform molar pregnancy in relation to maternal age. BJOG 2002; 109: 99102. 14 Garrett LA, Garner EI, Feltmate CM, Goldstein DP, Berkowitz RS. Subsequent pregnancy outcomes in patients with molar pregnancy and persistent gestational trophoblastic neoplasia. J Reprod Med 2008; 53: 48186. 15 Sebire NJ, Fisher RA, Foskett M, Rees H, Seckl MJ, Newlands ES. Risk of recurrent hydatidiform mole and subsequent pregnancy outcome following complete or partial hydatidiform molar pregnancy. BJOG 2003; 110: 2226. 16 Tuncer ZS, Bernstein MR, Wang J, Goldstein DP, Berkowitz RS. Repetitive hydatidiform mole with dierent male partners. Gynecol Oncol 1999; 75: 22426. 17 Palmieri C, Fisher RA, Sebire NJ, et al. Placental site trophoblastic tumour arising from a partial hydatidiform mole. Lancet 2005; 366: 688. 18 Seckl MJ, Fisher RA, Salerno GA, et al. Choriocarcinoma and partial hydatidiform moles. Lancet 2000; 356: 3639.

Conclusions
Although outcomes for more than 98% of women with gestational trophoblastic neoplasia are excellent, a few women die from the disease, mainly because of late presentation and diagnosis or drug resistance. Consequently, novel therapies with improved ecacy and reduced toxic eects need to be identied. Additionally, after diagnosis of molar pregnancy, women have to wait many weeks or months to nd out whether they need chemotherapy. Therefore, a new prognostic test at the time of initial molar diagnosis is needed to identify those who might develop malignant disease. Of course, if residual complete hydatidiform mole acquires additional genetic changes after evacuation, such a test might never be possible. However, evidence comparing early versus delayed evacuation of complete hydatidiform mole alone and in pregnancies with a healthy twin suggest that delayed termination has no increased risk of malignant disease.76,84 This nding suggests that hydatidiform moles are probably pre-programmed to behave malignantly at an early stage of development and before uterine evacuation. The optimum management strategy for women with repetitive molar pregnancies to allow healthy pregnancy is unknown; even though the causative gene has been
726

www.thelancet.com Vol 376 August 28, 2010

Seminar

19 20

21

22

23

24

25 26

27 28

29

30 31

32

33

34

35

36

37

38

39

40

41

Ngan S, Seckl MJ. Gestational trophoblastic neoplasia management: an update. Curr Opin Oncol 2007; 19: 48691. Schmid P, Nagai Y, Agarwal R, et al. Prognostic markers and longterm outcome of placental-site trophoblastic tumours: a retrospective observational study. Lancet 2009; 374: 4855. Buckley JD, Henderson BE, Morrow CP, Hammond CB, Kohorn EI, Austin DF. Case-control study of gestational choriocarcinoma. Cancer Res 1988; 48: 100410. Palmer JR, Driscoll SG, Rosenberg L, et al. Oral contraceptive use and risk of gestational trophoblastic tumors. J Natl Cancer Inst 1999; 91: 63540. Stone M, Dent J, Kardana A, Bagshawe KD. Relationship of oral contraceptive to development of trophoblastic tumour after evacuation of hydatidiform mole. Br J Obstet Gynaecol 1976; 86: 91316. Costa HL, Doyle P. Inuence of oral contraceptives in the development of post-molar trophoblastic neoplasiaa systematic review. Gynecol Oncol 2006; 100: 57985. Kajii T, Ohama K. Androgenetic origin of hydatidiform mole. Nature 1977; 268: 63334. Yamashita K, Ishikawa M, Shimizu T, Kuroda M. HLA antigens in husband-wife pairs with trophoblastic tumour. Gynaecol Oncol 1981; 12: 6874. Fisher RA, Newlands ES. Gestational trophoblastic disease: molecular and genetic studies. J Reprod Med 1998; 43: 8197. Pattillo RA, Sasaki S, Katayama KP, Roesler M, Mattingly RF. Genesis of 46,XY hydatidiform mole. Am J Obstet Gynecol 1981; 141: 10405. Azuma C, Saji F, Tokugawa Y, et al. Application of gene amplication by polymerase chain reaction to genetic analysis of molar mitochondrial DNA: the detection of anuclear empty ovum as the cause of complete mole. Gynecol Oncol 1991; 40: 2933. Fisher RA, Hodges MD, Newlands ES. Familial recurrent hydatidiform mole: a review. J Reprod Med 2004; 49: 595601. Moglabey YB, Kircheisen R, Seoud M, El Mogharbel N, Van den Veyver I, Slim R. Genetic mapping of a maternal locus responsible for familial hydatidiform moles. Hum Mol Genet 1999; 8: 66771. Murdoch S, Djuric U, Mazhar B, et al. Mutations in NALP7 cause recurrent hydatidiform moles and reproductive wastage in humans. Nat Genet 2006; 38: 30002. Kou YC, Shao L, Peng HH, et al. A recurrent intragenic genomic duplication, other novel mutations in NLRP7 and imprinting defects in recurrent biparental hydatidiform moles. Mol Hum Reprod 2008; 14: 3340. Wang CM, Dixon PH, Decordova S, et al. Identication of 13 novel NLRP7 mutations in 20 families with recurrent hydatidiform mole; missense mutations cluster in the leucine-rich region. J Med Genet 2009; 46: 56975. Deveault C, Qian JH, Chebaro W, et al. NLRP7 mutations in women with diploid androgenetic and triploid moles: a proposed mechanism for mole formation. Hum Mol Genet 2009; 18: 88897. Szulman A, Surti U. The syndromes of hydatidiform mole. I. Cytogenetic and morphological correlations. Am J Obstet Gynaecol 1978; 13: 66571. Lawler SD, Fisher RA, Dent J. A prospective genetic study of complete and partial hydatidiform moles. Am J Obstet Gynecol 1991; 164: 127077. Lage JM, Mark SD, Roberts DJ, Goldstein DP, Bernstein MR, Berkowitz RS. A ow cytometric study of 137 fresh hydropic placentas: correlation between types of hydatidiform moles and nuclear DNA ploidy. Obstet Gynecol 1992; 79: 40310. Genest DR, Ruiz RE, Weremowicz S, Berkowitz RS, Goldstein DP, Dorfman DM. Do nontriploid partial hydatidiform moles exist? A histologic and ow cytometric reevaluation of nontriploid specimens. J Reprod Med 2002; 47: 36368. Sebire NJ, Fisher RA, Rees HC. Histopathological diagnosis of partial and complete hydatidiform mole in the rst trimester of pregnancy. Pediatr Dev Pathol 2003; 6: 6977. Sebire NJ, Makrydimas G, Agnantis NJ, Zagorianakou N, Rees H, Fisher RA. Updated diagnostic criteria for partial and complete hydatidiform moles in early pregnancy. Anticancer Res 2003; 23: 172328.

42 43

44

45

46

47

48 49

50

51

52

53 54

55

56

57

58

59

60

61 62 63

64

Sebire NJ, Seckl MJ. Gestational trophoblastic disease: current management of hydatidiform mole. BMJ 2008; 337: 1193. Castrillon DH, Sun D, Weremowicz S, Fisher RA, Crum CP, Genest DR. Discrimination of complete hydatidiform mole from its mimics by immunohistochemistry of the paternally imprinted gene product p57kip2. Am J Surg Pathol 2001; 25: 122530. Fisher RA, Hodges MD, Rees HC, et al. The maternally transcribed gene p57(KIP2) (CDNK1C) is abnormally expressed in both androgenetic and biparental complete hydatidiform moles. Hum Mol Genet 2002; 11: 326772. Fisher RA, Newlands ES. Rapid diagnosis and classication of hydatidiform moles with polymerase chain reaction. Am J Obstet Gynecol 1993; 168: 56369. Makrydimas G, Sebire NJ, Thornton SE, Zagorianakou N, Lolis D, Fisher RA. Complete hydatidiform mole and normal live birth: a novel case of conned placental mosaicism: case report. Hum Reprod 2002; 17: 245963. Osada H, Iitsuka Y, Matsui H, Sekiya S. A complete hydatidiform mole coexisting with a normal fetus was conrmed by variable number tandem repeat (VNTR) polymorphism analysis using polymerase chain reaction. Gynecol Oncol 1995; 56: 9093. Fox H, Sebire NJ. Pathology of the placenta, 3rd edn. Philadelphia: Elsevier, 2007. Sebire NJ, Lindsay I, Fisher RA, Seckl MJ. Intraplacental choriocarcinoma: experience from a tertiary referral center and relationship with infantile choriocarcinoma. Fetal Pediatr Pathol 2005; 24: 2129. Baergen RN, Rutgers JL, Young RH, Osann K, Scully RE. Placental site trophoblastic tumor: a study of 55 cases and review of the literature emphasizing factors of prognostic signicance. Gynecol Oncol 2006; 100: 51120. Shih IM, Kurman RJ. Epithelioid trophoblastic tumor: a neoplasm distinct from choriocarcinoma and placental site trophoblastic tumor simulating carcinoma. Am J Surg Pathol 1998; 22: 1393403. Mitchell H, Seckl MJ. Discrepancies between commercially available immunoassays in the detection of tumour-derived hCG. Mol Cell Endocrinol 2007; 260: 31013. Cole LA. Human chorionic gonadotropin tests. Expert Rev Mol Diagn 2009; 9: 72147. Sturgeon CM, Berger P, Bidart JM, et al. Dierences in recognition of the 1st WHO international reference reagents for hCG-related isoforms by diagnostic immunoassays for human chorionic gonadotropin. Clin Chem 2009; 55: 148491. Cole LA, Shahabi S, Butler SA, et al. Utility of commonly used commercial human chorionic gonadotropin immunoassays in the diagnosis and management of trophoblastic diseases. Clin Chem 2001; 47: 30815. Cole LA, Butler SA, Khanlian SA, et al. Gestational trophoblastic diseases: 2. Hyperglycosylated hCG as a reliable marker of active neoplasia. Gynecol Oncol 2006; 102: 15159. Palmieri C, Dhillon T, Fisher RA, et al. Management and outcome of healthy women with a persistently elevated beta-hCG. Gynecol Oncol 2007; 106: 3543. Cole LA, Khanlian SA, Muller CY, Giddings A, Kohorn E, Berkowitz R. Gestational trophoblastic diseases: 3. Human chorionic gonadotropin-free beta-subunit, a reliable marker of placental site trophoblastic tumors. Gynecol Oncol 2006; 102: 16064. Harvey RA, Pursglove HD, Schmid P, Savage PM, Mitchell HD, Seckl MJ. Human chorionic gonadotropin free beta-subunit measurement as a marker of placental site trophoblastic tumors. J Reprod Med 2008; 53: 64348. Hou JL, Wan XR, Xiang Y, Qi QW, Yang XY. Changes of clinical features in hydatidiform mole: analysis of 113 cases. J Reprod Med 2008; 53: 62933. Berkowitz RS, Goldstein DP, Bernstein MR. Natural history of partial molar pregnancy. Obstet Gynecol 1985; 66: 67781. Czernobilsky B, Barash A, Lancet M. Partial moles: a clinicopathologic study of 25 cases. Obstet Gynecol 1982; 59: 7577. Santos-Ramos R, Forney JP, Schwarz BE. Sonographic ndings and clinical correlations in molar pregnancy. Obstet Gynecol 1980; 56: 18692. Reid MH, McGahan JP, Oi R. Sonographic evaluation of hydatidiform mole and its look-alikes. AJR Am J Roentgenol 1983; 140: 30711.

www.thelancet.com Vol 376 August 28, 2010

727

Seminar

65

66

67

68 69

70 71

72

73

74

75

76

77

78 79 80

81

82 83

84

85

86

87

88

Benson CB, Genest DR, Bernstein MR, Soto-Wright V, Goldstein DP, Berkowitz RS. Sonographic appearance of rst trimester complete hydatidiform moles. Ultrasound Obstet Gynecol 2000; 16: 18891. Crade M, Weber PR. Appearance of molar pregnancy 9.5 weeks after conception. Use of transvaginal ultrasound for early diagnosis. J Ultrasound Med 1991; 10: 47374. Fine C, Bundy AL, Berkowitz RS, Boswell SB, Berezin AF, Doubilet PM. Sonographic diagnosis of partial hydatidiform mole. Obstet Gynaecol 1989; 73: 41418. Jauniaux E, Nicolaides KH. Early ultrasound diagnosis and followup of molar pregnancies. Ultrasound Obstet Gynecol 1997; 9: 1721. Sherer DM, Allen T, Woods J Jr. Transvaginal sonographic diagnosis of a hydatidiform mole occurring two weeks after curettage for an incomplete abortion. J Clin Ultrasound 1991; 19: 22426. Woo JS, Wong LC, Hsu C, Ma HK. Sonographic appearances of the partial hydatidiform mole. J Ultrasound Med 1983; 2: 26164. Fowler DJ, Lindsay I, Seckl MJ, Sebire NJ. Routine pre-evacuation ultrasound diagnosis of hydatidiform mole: experience of more than 1000 cases from a regional referral center. Ultrasound Obstet Gynecol 2006; 27: 5660. Johns J, Greenwold N, Buckley S, Jauniaux E. A prospective study of ultrasound screening for molar pregnancies in missed miscarriages. Ultrasound Obstet Gynecol 2005; 25: 49397. Sebire NJ, Rees H, Paradinas F, Seckl M, Newlands E. The diagnostic implications of routine ultrasound examination in histologically conrmed early molar pregnancies. Ultrasound Obstet Gynecol 2001; 18: 66265. Fowler DJ, Lindsay I, Seckl MJ, Sebire NJ. Histomorphometric features of hydatidiform moles in early pregnancy: relationship to detectability by ultrasound examination. Ultrasound Obstet Gynecol 2007; 29: 7680. Royal College of Obstetricians and Gynaecologists. The management of early pregnancy loss. Green-top guideline. London: Royal College of Obstetricians and Gynaecologists, 2006: 118. Seckl MJ, Dhillon T, Dancey G, et al. Increased gestational age at evacuation of a complete hydatidiform mole: does it correlate with increased risk of requiring chemotherapy? J Reprod Med 2004; 49: 52730. Seckl MJ, Gillmore R, Foskett M, Sebire NJ, Rees H, Newlands ES. Routine terminations of pregnancyshould we screen for gestational trophoblastic neoplasia? Lancet 2004; 364: 70507. Berkowitz RS, Goldstein DP. Clinical practice. Molar pregnancy. N Engl J Med 2009; 360: 163945. Hancock BW, Tidy JA. Current management of molar pregnancy. J Reprod Med 2002; 47: 34754. Tidy JA, Hancock BW, Newlands ES. The management of gestational trophoblastic neoplasia. Clinical guideline 38. London: Royal College of Obstetricians and Gynaecologists, 2004. Xia ZF, Song HZ, Tang MY. Risk of malignancy and prognosis using a provisional scoring system in hydatidiform mole. Chin Med J (Engl) 1980; 93: 60512. Elias K, Goldstein DP, Berkowitz RS. Complete hydatidiform mole in women older than age 50. J Reprod Med 2009 (in press). Matsui H, Sekiya S, Hando T, Wake N, Tomoda Y. Hydatidiform mole coexistent with a twin live fetus: a national collaborative study in Japan. Hum Reprod 2000; 15: 60811. Sebire NJ, Foskett M, Paradinas FJ, et al. Outcome of twin pregnancies with complete hydatidiform mole and healthy co-twin. Lancet 2002; 359: 216566. Hancock BW, Nazir K, Everard JE. Persistent gestational trophoblastic neoplasia after partial hydatidiform mole incidence and outcome. J Reprod Med 2006; 51: 76466. Feltmate CM, Growdon WB, Wolfberg AJ, et al. Clinical characteristics of persistent gestational trophoblastic neoplasia after partial hydatidiform molar pregnancy. J Reprod Med 2006; 51: 90206. Sebire NJ, Foskett M, Short D, et al. Shortened duration of human chorionic gonadotrophin surveillance following complete or partial hydatidiform mole: evidence for revised protocol of a UK regional trophoblastic disease unit. BJOG 2007; 114: 76062. Kerkmeijer L, Wielsma S, Bekkers R, Pyman J, Tan J, Quinn M. Guidelines following hydatidiform mole: a reappraisal. Aust N Z J Obstet Gynaecol 2006; 46: 11218.

89

90

91

92

93

94

95 96

97 98

99

100

101 102

103

104

105

106

107

108

109

110

111

Wielsma S, Kerkmeijer L, Bekkers R, Pyman J, Tan J, Quinn M. Persistent trophoblast disease following partial molar pregnancy. Aust N Z J Obstet Gynaecol 2006; 46: 11923. Fisher RA, Soteriou BA, Meredith L, Paradinas FJ, Newlands ES. Previous hydatidiform mole identied as the causative pregnancy of choriocarcinoma following birth of normal twins. Int J Cancer 1995; 5: 6470. Fisher RA, Savage PM, MacDermott C, et al. The impact of molecular genetic diagnosis on the management of women with hCG-producing malignancies. Gynecol Oncol 2007; 107: 41319. Kohorn EI. Negotiating a staging and risk factor scoring system for gestational trophoblastic neoplasia. A progress report. J Reprod Med 2002; 47: 44550. Agarwal R, Strickland S, McNeish IA, et al. Doppler ultrasonography of the uterine artery and the response to chemotherapy in patients with gestational trophoblastic tumors. Clin Cancer Res 2002; 8: 114247. Lim A. Imaging gestational trophoblastic neoplasia. International Society for the Study of Trophoblastic Diseases XV World Congress on Gestational Trophoblastic Disease; Cochin, India; Nov 1215, 2009, 27 (abstr). Berkowitz RS, Goldstein DP. Current management of gestational trophoblastic diseases. Gynecol Oncol 2009; 112: 65462. Ngan HY, Chan FL, Au VW, Cheng DK, Ng TY, Wong LC. Clinical outcome of micrometastasis in the lung in stage IA persistent gestational trophoblastic disease. Gynecol Oncol 1998; 70: 19294. Darby S, Jolley I, Pennington S, Hancock BW. Does chest CT matter in the staging of GTN? Gynecol Oncol 2009; 112: 15560. Pezeshki M, Hancock BW, Silcocks P, et al. The role of repeat uterine evacuation in the management of persistent gestational trophoblastic disease. Gynecol Oncol 2004; 95: 42329. van Trommel NE, Massuger LF, Verheijen RH, Sweep FC, Thomas CM. The curative eect of a second curettage in persistent trophoblastic disease: a retrospective cohort survey. Gynecol Oncol 2005; 99: 613. Garner EI, Feltmate CM, Goldstein DP, Berkowitz RS. The curative eect of a second curettage in persistent trophoblastic disease: a retrospective cohort survey. Gynecol Oncol 2005; 99: 35. Savage P, Seckl MJ. The role of repeat uterine evacuation in trophoblast disease. Gynecol Oncol 2005; 99: 25152. Alazzam M, Tidy J, Hancock BW, Osborne R. First line chemotherapy in low risk gestational trophoblastic neoplasia. Cochrane Database Syst Rev 2009; 1: CD007102. Kohorn EI. Is lack of response to single-agent chemotherapy in gestational trophoblastic disease associated with dose scheduling or chemotherapy resistance? Gynecol Oncol 2002; 85: 3639. Osborne R, Filiaci V, Schink JC. A randomized phase III trial comparing weekly parenteral methorexate and pulsed dactinomycin as primary management of low-risk gestational trophoblastic neoplasia: a Gynecologic Oncology Group study. Gynecol Oncol 2008; 108: S2S3. McNeish IA, Strickland S, Holden L, et al. Low risk persistent gestational trophoblastic disease: outcome following initial treatment with low-dose methotrexate and folinic acid, 19922000. J Clin Oncol 2002; 20: 183844. Berkowitz RS, Goldstein DP, Bernstein MR. Ten years experience with methotrexate and folinic acid as primary therapy for gestational trophoblastic disease. Gynecol Oncol 1986; 23: 11118. Growdon WB, Wolfberg AJ, Goldstein DP, et al. Evaluating methotrexate treatment in patients with low-risk postmolar gestational trophoblastic neoplasia. Gynecol Oncol 2009; 112: 35357. McGrath S, Short D, Harvey R. The management and outcome of women with post-hydatidiform mole low-risk gestational trophoblastic neoplasia, but hCG levels in excess of 100 000 IU L. Br J Cancer 2010; 102: 81014. Bagshawe KD, Lawler SD, Paradinas FJ, Dent J, Brown P, Boxer GM. Gestational trophoblastic tumours following initial diagnosis of partial hydatidiform mole. Lancet 1990; 335: 107476. Newlands ES, Holden L, Seckl MJ, McNeish I, Strickland S, Rustin GJ. Management of brain metastases in patients with high-risk gestational trophoblastic tumors. J Reprod Med 2002; 47: 46571. Seckl MJ, Rustin GJS, Newlands ES, Gwyther SJ, Bomanji J. Pulmonary embolism, pulmonary hypertension, and choriocarcinoma. Lancet 1991; 338: 131315.

728

www.thelancet.com Vol 376 August 28, 2010

Seminar

112 Athanassiou A, Begent RHJ, Newlands ES, Parker D, Rustin GJS, Bagshawe KD. Central nervous system metastases of choriocarcinoma: 23 years experience at Charing Cross Hospital. Cancer 1983; 52: 172835. 113 Patten DK, Lindsay I, Fisher R, Sebire N, Savage PM, Seckl MJ. Gestational choriocarcinoma mimicking a uterine adenocarcinoma. J Clin Oncol 2008; 26: 512627. 114 Bagshawe KD. Treatment of trophoblastic tumours. Recent results Cancer Res 1977; 62: 19299. 115 Deng L, Yan X, Zhang J, Wu T. Combination chemotherapy for highrisk gestational trophoblastic tumour. Cochrane Database Syst Rev 2009; 3: CD005196. 116 Newlands ES, Bagshawe KD, Begent RH, Rustin GJ, Holden L. Results with the EMA/CO (etoposide, methotrexate, actinomycin D, cyclophosphamide, vincristine) regimen in high risk gestational trophoblastic tumours, 1979-1989. Br J Obstet Gynaecol 1991; 98: 55057. 117 Kim SJ, Bae SN, Kim JH, et al. Eects of multiagent chemotherapy and independent risk factors in the treatment of high-risk GTT25 years experiences of KRI-TRD. Int J Gynaecol Obstet 1998; 60 (suppl 1): S8596. 118 Bolis G, Bonazzi C, Landoni F, et al. EMA/CO regimen in high-risk gestational trophoblastic tumor (GTT). Gynecol Oncol 1988; 31: 43944. 119 Bower M, Newlands ES, Holden L, et al. EMA/CO for high-risk gestational trophoblastic tumours: results from a cohort of 272 patients. J Clin Oncol 1997; 15: 263643. 120 Lu WG, Ye F, Shen YM, et al. EMA-CO chemotherapy for high-risk gestational trophoblastic neoplasia: a clinical analysis of 54 patients. Int J Gynecol Cancer 2008; 18: 35762. 121 Turan T, Karacay O, Tulunay G, et al. Results with EMA/CO (etoposide, methotrexate, actinomycin D, cyclophosphamide, vincristine) chemotherapy in gestational trophoblastic neoplasia. Int J Gynecol Cancer 2006; 16: 143238. 122 Crawford RA, Newlands E, Rustin GJ, Holden L, AHern R, Bagshawe KD. Gestational trophoblastic disease with liver metastases: the Charing Cross experience. Br J Obstet Gynaecol 1997; 104: 10509. 123 Seckl MJ, Ahamed E, Short D, North B, Schmid P, Savage P. Survival of women with gestational trophoblastic neoplasia and liver metastases: is it improving? International Society for the Study of Trophoblastic Diseases XV World Congress on Gestational Trophoblastic Disease; Cochin, India; Nov 1215, 2009. 8990 (abstr). 124 Powles T, Young A, Sammit A, et al. The signicance of the time interval between antecedent pregnancy and diagnosis of high-risk gestational trophoblastic tumours. Br J Cancer 2006; 95: 114547. 125 Seckl MJ, Newlands ES. Chapter 12. Investigation and treatment of patients with persistent gestational trophoblastic disease and gestational trophoblastic tumours/neoplasia in the United Kingdom. In: Hancock BW, Seckl MJ, Berkowitz RS, Cole LA, eds. Gestational trophoblastic disease, 3rd edn. http://www.isstd.org/isstd/book.html (accessed Nov 20, 2009). 126 Powles T, Savage P, Short D, Young A, Pappin C, Seckl MJ. Residual lung lesions after completion of chemotherapy for gestational trophoblastic neoplasia: should we operate? Br J Cancer 2006; 94: 5154. 127 Yang J, Xiang Y, Wan X, Yang X. The prognosis of gestational trophoblastic neoplasia patient with residual lung tumor after completing treatment. Gynecol Oncol 2006; 103: 47982. 128 Powles T, Savage PM, Stebbing J, et al. A comparison of patients with relapsed and chemo-refractory gestational trophoblastic neoplasia. Br J Cancer 2007; 96: 73237. 129 Dhillon T, Palmieri C, Sebire NJ, et al. Value of whole body FDG-PET to identify the active site of gestational trophoblastic neoplasia. J Reprod Med 2006; 51: 87987. 130 Newlands ES. The management of recurrent and drug-resistant gestational trophoblastic neoplasia (GTN). Best Pract Res Clin Obstet Gynaecol 2003; 17: 90523. 131 Lurain JR, Nejad B. Secondary chemotherapy for high-risk gestational trophoblastic neoplasia. Gynecol Oncol 2005; 97: 61823. 132 Newlands ES, Mulholland PJ, Holden L, Seckl MJ, Rustin GJ. Etoposide and cisplatin/etoposide, methotrexate, and actinomycin D (EMA) chemotherapy for patients with high-risk gestational trophoblastic tumors refractory to EMA/cyclophosphamide and vincristine chemotherapy and patients presenting with metastatic placental site trophoblastic tumors. J Clin Oncol 2000; 18: 85459.

133 Pandian Z, Seckl MJ, Smith R, Lees DA. Gestational choriocarcinoma: an unusual presentation with response to gemcitabine and surgery. BJOG 2004; 111: 38284. 134 Jones WB, Schneider J, Shapiro F, Lewis JLJ. Treatment of resistant gestational choriocarcinoma with taxol: a report of two cases. Gynaecol Oncol 1996; 61: 12630. 135 Osborne R, Covens A, Mirchandani D, Gerulath A. Successful salvage of relapsed high-risk gestational trophoblastic neoplasia patients using a novel paclitaxel-containing doublet. J Reprod Med 2004; 49: 65561. 136 Termrungruanglert W, Kudelka AP, Piamsomboon S, et al. Remission of refractory gestational trophoblastic disease with high-dose paclitaxel. Anticancer Drugs 1996; 7: 50306. 137 Shorbagi A, Aksoy S, Kilickap S, Guler N. Successful salvage therapy of resistant gestational trophoblastic disease with ifosfamide and paclitaxel. Gynecol Oncol 2005; 97: 72223. 138 Wang J, Short D, Sebire NJ, et al. Salvage chemotherapy of relapsed or high-risk gestational trophoblastic neoplasia (GTN) with paclitaxel/cisplatin alternating with paclitaxel/etoposide (TP/TE). Ann Oncol 2008; 19: 157883. 139 El-Helw LM, Seckl MJ, Haynes R, et al. High-dose chemotherapy and peripheral blood stem cell support in refractory gestational trophoblastic neoplasia. Br J Cancer 2005; 93: 62021. 140 Beyer J, Kramar A, Mandanas R, et al. High-dose chemotherapy as salvage treatment in germ cell tumors: a multivariate analysis of prognostic variables. J Clin Oncol 1996; 14: 263845. 141 McNeish IA, Kanfer EJ, Haynes R, et al. Paclitaxel-containing highdose chemotherapy for relapsed or refractory testicular germ cell tumours. Br J Cancer 2004; 90: 116975. 142 Feltmate CM, Genest DR, Goldstein DP, Berkowitz RS. Advances in the understanding of placental site trophoblastic tumor. J Reprod Med 2002; 47: 33741. 143 Papadopoulos AJ, Foskett M, Seckl MJ, et al. Twenty-ve years clinical experience of placental site trophoblastic tumors. J Reprod Med 2002; 47: 46064. 144 Newlands ES, Bower M, Fisher RA, Paradinas FJ. Management of placental site trophoblastic tumours. J Reprod Med 1998; 43: 5359. 145 Pfeer PE, Sebire N, Lindsay I, McIndoe A, Lim A, Seckl MJ. Fertility-sparing partial hysterectomy for placental-site trophoblastic tumour. Lancet Oncol 2007; 8: 74446. 146 Leiserowitz GS, Webb MJ. Treatment of placental site trophoblastic tumor with hysterotomy and uterine reconstruction. Obstet Gynecol 1996; 88: 69699. 147 Machtinger R, Gotlieb WH, Korach J, et al. Placental site trophoblastic tumor: outcome of ve cases including fertility preserving management. Gynecol Oncol 2005; 96: 5661. 148 Tsuji Y, Tsubamoto H, Hori M, Ogasawara T, Koyama K. Case of PSTT treated with chemotherapy followed by open uterine tumor resection to preserve fertility. Gynecol Oncol 2002; 87: 30307. 149 Bower M, Rustin GJS, Newlands ES, et al. Chemotherapy for gestational trophoblastic tumours hastens menapause by 3 years. Eur J Cancer 1998; 34: 120407. 150 Woolas RP, Bower M, Newlands ES, Seckl MJ, Short D, Holden L. Inuence of chemotherapy for gestational trophoblastic disease on subsequent pregnancy outcome. Br J Obset Gynaecol 1998; 105: 103235. 151 Rustin GJS, Booth M, Dent J, Salt S, Rustin F, Bagshawe KD. Pregnancy after cytotoxic chemotherapy for gestational trophoblastic tumours. Br Med J 1984; 288: 10306. 152 Blagden SP, Foskett MA, Fisher RA, et al. The eect of early pregnancy following chemotherapy on disease relapse and foetal outcome in women treated for gestational trophoblastic tumours. Br J Cancer 2002; 86: 2630. 153 Rustin GJS, Newlands ES, Lutz J-M, et al. Combination but not single agent methotrexate chemotherapy for gestational trophoblastic tumours (GTT) increases the incidence of second tumours. J Clin Oncol 1996; 14: 276973.

www.thelancet.com Vol 376 August 28, 2010

729