03 (2006) 642–648www.elsevier.com/locate/ygyno

Gynecologic Oncology 1

Evaluation of deletions in 7q11.2 and 8p12–p21 as prognostic indicators oftumour development following molar pregnancy

Beverley Burke a, Neil J. Sebire b, Jill Moss b, Matthew D. Hodges a, Michael J. Seckl a,Edward S. Newlands a, Rosemary A. Fisher a,⁎

a Department of Oncology, Division of Surgery, Oncology, Reproductive Biology and Anaesthetics, Imperial College London, Charing Cross Hospital,Fulham Palace Road, London W6 8RF, UK

b Department of Histopathology, Charing Cross Hospital, London W6 8RF, UK

Received 5 December 2005Available online 27 June 2006

Abstract

Objectives. Previous studies have identified loss of chromosomal regions 7p12–q11.2 and 8p12–p21 in choriocarcinoma suggesting thatsuppressor genes involved in tumour development may be located within these regions. Our objectives were to refine the regions of loss andevaluate these deletions as prognostic indicators of trophoblastic tumour development following molar pregnancy.

Methods. Fluorescent microsatellite genotyping was used to perform deletion mapping in a series of thirty-nine gestational trophoblastictumours (GTT) including both choriocarcinoma and placental site trophoblastic tumours.

Results. Significant loss of heterozygosity (LOH) was found for both regions in GTT that originated in non-molar pregnancies. Although nocommon interval of loss was found in those GTTwith LOH for the 7q11.2 region, for the 8p12–p21 locus, markers D8S1731 and NEFL defined aminimal region of loss in all tumours showing LOH. However, complete LOH of either region occurred in only a minority of tumours (20%;chromosome 7: 24%; chromosome 8) suggesting that loss of neither region is likely to be a primary event in the development of GTT. This wasfurther supported by the observation that no deletions were found in either region for the fourteen GTT that followed complete molar pregnancies.

Conclusions. While we have defined a minimal interval in 8p12–p21 in which tumour suppressor genes involved in GTT are likely to belocated, the data suggest that deletions in 7q11.2 or 8p12–p21 are unlikely to be useful prognostic indicators in the management of patients withmolar pregnancies.© 2006 Elsevier Inc. All rights reserved.

Keywords: Gestational trophoblastic tumour; Hydatidiform mole; Chromosome 7q; Chromosome 8p

Introduction

Gestational trophoblastic tumours (GTT) are an unusualgroup of neoplasms that originate not from a patient's owntissues, but from the placental tissue of a conceptus. AlthoughGTT may occur after any type of pregnancy they are in the orderof 2000 times more likely to occur following a hydatidiformmole (HM) than a non-molar pregnancy [1]. In the UK, molarpregnancies occur about once in every 700 live births [2] andapproximately 10% of women who have a HM subsequentlyrequire chemotherapy for a GTT. This may be an invasive mole,

⁎ Corresponding author. Fax: +44 208 748 5665.E-mail address: r.fisher@imperial.ac.uk (R.A. Fisher).

0090-8258/$ - see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.ygyno.2006.04.015

malignant choriocarcinoma or placental site trophoblastictumour (PSTT). Consequently, following a molar pregnancy,women are monitored with serial measurements of humanchorionic gonadotrophin (hCG) to detect persistent trophoblas-tic disease (PTD). A number of molecular markers have beenshown to play some role in the development of PTD [3] but asyet none are able to accurately predict whether HM will resolvespontaneously or persist following uterine evacuation. Geneticmarkers that can be used to identify HM that will progress toGTT would greatly facilitate the appropriate management ofwomen with molar pregnancies.

In many other types of tumour, development of themalignant phenotype is a multistep process associated with aseries of genetic changes. These changes may be identified as

Table 1Clinicopathological summary of gestational trophoblastic tumour cases

Casenumber

Mean age atdiagnosis

Causativepregnancy

Pathology Site oftumour

1 25 CHMa IMb Uterus2–9 31 (22–45) CHM CCc Uterus10–11 29 (28–29) CHM CC Lung12 30 CHM CC Lymph

Node13–14 34 (30–38) CHM PSTTd Uterus15–22 32 (23–43) Non-molar CC Uterus23–25 29 (25–31) Non-molar CC Ovary26 21 Non-molar CC Fallopian

tube27–32 34 (27–45) Non-molar CC Lung33 25 Non-molar CC Lymph node34–39 33 (25–44) Non-molar PSTT Uterusa CHM, complete hydatidiform mole.b IM, invasive mole.c CC, choriocarcinoma.d PSTT, placental site trophoblastic tumour.

643B. Burke et al. / Gynecologic Oncology 103 (2006) 642–648

loss of heterozygosity (LOH) or amplification of specificregions of the genome and are likely to indicate the location oftumour suppressor genes or oncogenes, respectively. Refine-ment of these regions in different tumour types can lead tocharacterisation of the genes involved in the malignant pathwayfor that particular tumour. Although it is known that completehydatidiform mole (CHM) is more likely to progress to a GTTthan any other type of pregnancy, GTT themselves are poorlycharacterised genetically since patients with these tumours aregenerally treated with chemotherapy following clinical diagno-sis and tumour tissue is therefore rarely available for study.Where cytogenetic studies have been performed, choriocarci-noma have been found to exhibit a range of abnormalities,including chromosomal gains, losses and rearrangements but noconsistent changes [4–9]. More recent molecular genetic studieshave, however, demonstrated frequent loss of specific regions ofthe genome. The most significant of these are deletions of7p12–q11.2 [10], amplification of 7q21–q31 and loss of 8p12–p21 [11]. Allelic losses of 7p12–q11.2 and 8p12–p21, that arelikely to indicate the sites of tumour suppressor genes involvedin the development of choriocarcinoma, have previously beenproposed as the sites of putative tumour suppressor genes inseveral other malignancies including ovarian and breast cancer[12,13].

Using microsatellite genotyping, LOH should be readilyidentified in GTT that follow non-molar pregnancies since thesetumours are biallelic. However, LOH may be masked in post-molar GTT since the CHM from which they originate usuallyarise from a duplicated sperm and are consequently homozy-gous [14]. Despite this, Matsuda et al. [10] observed deletionsof 7p12–q11.2 in post-mole choriocarcinoma. Since thesedeletions were homozygous and occurred in 70% of post-moletumours we hypothesised that in these cases loss of geneticmaterial may have been present prior to duplication of the spermand CHM development. Loss of the 7p12–q11.2 region mighttherefore be an early marker of CHM likely to progress to GTT.

In this study, we used deletion mapping to assess thefrequency and refine the interval of chromosomal loss in 7q11.2and 8p12–p21 for a series of choriocarcinoma and PSTT thatdeveloped from both molar and non-molar pregnancies. Morespecifically we sought to determine whether loss of either ofthese regions might be a useful prognostic marker with which toidentify those CHM at risk of progressing to a GTT.

Materials and methods

Patient material

Trophoblastic tumours in which a gestational origin had been confirmed bygenetic diagnosis were identified from the trophoblastic tumour database atCharing Cross Hospital. Forty-eight cases of GTT were identified for whichblood samples were also available from both parents. Fourteen had arisen from aCHM (Table 1). In seven, the clinically antecedent pregnancy was a CHM. Infour the GTT followed a non-molar pregnancy but was shown to have originatedin an earlier CHMwhile in three the tumour originated in an unrecognised molarpregnancy. Two cases of GTT originating in CHM were excluded from thestudy, as DNA from the tumour could not be amplified reliably. Genetic analysisalso demonstrated that all post-mole tumours in the study had originated inCHM of monospermic origin. Twenty-five tumours followed non-molar

pregnancies, four cases in which DNA failed to amplify reproducibly beingexcluded.

Thirty tumours were pathologically classified as choriocarcinoma, eight asPSTT and one as an invasive mole (Table 1). All PSTT examined were primarytumours from uterine curettings or hysterectomy specimens. Amongst thechoriocarcinoma, three of the 11 post-mole tumours and 11 of the 19 casesfollowing non-molar pregnancies were metastatic specimens from various sites(Table 1). The study was approved by the Riverside Research Ethics Committee.

Preparation of DNA

DNA was extracted from the blood of patients and their partners usingestablished protocols. For four cases DNA was available from fresh tumourtissue. For all other cases, DNAwas prepared from tissue microdissected fromformalin-fixed, paraffin wax-embedded pathological sections. In twenty-threecases, where tumour samples were small or the cells closely infiltrating maternaltissue, dissection of pure populations of tumour cells was facilitated by the useof laser capture microdissection (LCM) using the PixCell II LCM System(Arcturus, California, USA). DNA was prepared by digesting microdissectedtissue in 20 μl of digestion buffer (50 mM Tris (pH 8.5), 1 mM EDTA, 0.5%Tween 20) containing 200 μg/ml of proteinase K overnight at 37°C followed byheat inactivation of the proteinase K at 95°C for 8 min.

DNA prepared from archival material may show varying degrees ofdegradation. The integrity of each sample was verified by amplification of DNAfrom maternal tissue in the same sections with a panel of fluorescentmicrosatellite markers previously shown to be informative in performingdiagnosis. Following amplification, the genotype of the maternal tissue wasdetermined using ABI 310 PRISM GeneScan software (Applied Biosystems,Warrington, UK) as previously described [15] and compared with that of thepatient.

Deletion mapping of the 7q11.2 and 8p12–p21 regions

Deletion mapping was performed by fluorescent microsatellite genotypingof parental and tumour DNAwith a panel of markers for the critical regions inchromosome 7 and 8 previously described [10,11]. For chromosome 7 a panel offour markers, located in a 2.5 Mb region of chromosome 7p11.2 (Ensemblassembly 33; http://ensembl.org) were used to investigate deletions in thisregion (Table 2). These included D7S520 and D7S663, markers that delineatethe smallest common deletion previously described, and D7S502 and D7S482,the markers within this region that were most frequently lost [10]. MarkersD7S519, located in 7p13 and D7S669, located in 7q21 flank the centromeric andtelomeric ends of the critical region and were used as control markers. Allmarkers were tested on DNA prepared from maternal tissue in the same archival

Table 2Microsatellite markers used for deletion mapping

Microsatellite marker Primer sequences Allele size range (base pairs) Heterozygosity

ControlsD7S519 F a-ACAGCCAAGCATTTCTGCTG Rb-ACAGACCAGGACTCAGCCAG 256–268 0.82D7S699 F-ATGCAACCTACCCTCAAATG R-TACGGTTTACCCACATTGCTAT 179–189 0.807q11.2D7S520 F-CAACAGGTCCAGGCTATGTC R-TATCCATACACACCATGCCA 79–97 0.70CA52158 F-GGCAGTAATCTGGCTGGAT R-GGTGACCAGTAAATGTTTTTCTGA 129–158 0.73D7S502 F-GGAAGGTATGTTGCGG R-CCTTCCCTATCTGGAAGTTA 223–231 0.85D7S482 F-TGGCTCACGTCTGGAA R-TGTGGCTGGCTAATTTG 166–198 0.748p12–p21D8S1731 F-GCAGAATCTGTGGCTTTTGGT R-CCAACCAACCACCCAATCAA 112–136 0.84D8S1786 F-CTCCCAGCTTCCCGCAAA R-GTTTCCACACCGAAGCC 104–124 0.88NEFL F-GCAGTAGTGCCGCAGTTTCA R-TGCAATTCATCTTCCTTTCT 137–147 0.81D8S2611 F-GGCTTCTGCTGGCTACATA R-CCAGCTCATGTGGATGCTTA 140–168 0.90

a F, forward primer.b R, reverse primer.

644 B. Burke et al. / Gynecologic Oncology 103 (2006) 642–648

blocks as the tumour tissue to ensure efficient amplification. The microsatellitemarker D7S663 produced an amplicon greater than 300 base pairs in size thatfailed to amplify reliably from archival DNA and was therefore replaced with anovel, smaller microsatellite marker, CA52158 from the same region (Table 2).

Primers for four microsatellite markers D8S1731, D8S1786 (D8S136),NEFL and D8S2611, spanning the 13 Mb critical region previously described[11] and having heterozygosity values of greater than 0.7 (Human GenomeDatabase. http://www.gdb.org) were used to investigate the 8p12–p21 region inparental and tumour DNA (Table 2).

In most investigations of chromosomal deletions, tumour DNA is compareddirectly with that of the host in order to identify LOH in the tumour. For GTT,the genotype of the tumour reflects that of the causative pregnancy rather thanthe host. The genotype of the tumour was therefore compared to that of thepatient and the appropriate partner. In one case for which the partner was notavailable, the interval between the causative pregnancy and the diagnosis ofGTTwas greater than 18 years and DNA from the tumour was compared directlywith DNA from the daughter of the causative pregnancy.

Tumours that followed non-molar pregnancies were scored as showingcomplete LOH for a marker where only a single allele from one parent waspresent. Where a tumour was homozygous for an allele found in both parents,that marker was scored as uninformative. Most microsatellite markers used inthe investigation were dinucleotide repeats that show stutter bands andpreferential amplification of smaller alleles. Where a tumour had both amaternal and paternal contribution to the genome, resulting in a differentgenotype to either parent the relative height of the two alleles was comparedwith that found in control samples of the same genotype. A small number oftumours were found to show allelic imbalance rather than complete loss of oneallele. In these cases, the ratio of the height of the two alleles was calculated forthe average of three normal controls of the same genotype (N1/N2) and for thetumour (T1/T2). Allelic imbalance was calculated from the ratio of tumoursignal to that of the normal signal (T1/T2 over N1/N2). Ratios of <0.67 or >1.35were taken to indicate partial LOH for that locus. The genotypes of tumoursfollowing CHM were compared with that of the relevant paternal DNA toinvestigate the presence of homozygous deletions.

Results

Deletion mapping in GTT following non-molar pregnancies

None of the 25 GTT following non-molar pregnanciesshowed LOH for the control marker D7S519 (Fig. 1). Noneshowed complete LOH for the second control marker D7S669.However, one case of metastatic choriocarcinoma and onePSTT showed allelic imbalance in that the relative proportion of

the maternal to paternal allele were between 50 and 60% of thatexpected (Fig. 1), suggesting loss of the maternal allele at thislocus in a proportion of the tumour cells.

Deletion mapping of 7q11.2 identified complete LOH of oneor more markers in this region significantly more frequentlythan control loci (5 of 25 versus 0 of 25, respectively; z = 2.4,P = 0.025). Complete loss of one allele was observed for one ormore markers in three primary choriocarcinoma (Fig. 2) and onelung metastases. In addition one metastatic choriocarcinomaand one PSTT showed allelic imbalance for one or moremarkers suggesting partial LOH. One PSTT showing completeLOH for marker D7S482 also demonstrated partial LOH fortwo further markers (Fig. 3). Most frequent LOH was seen formarker D7S520 showing complete LOH in three and partialLOH in two of 17 (29%) informative cases.

Significant LOH was also found for markers in 8p12–21,compared with the control loci, five choriocarcinoma and onePSTT showing complete LOH (6 of 25 versus 0 of 25,respectively; z = 2.61, P = 0.011). LOH for chromosome 8 wasmore extensive than that observed for the 7q11.2 region. Mosttumours with deletions showed LOH for at least three markerswith five tumours potentially showing LOH for the whole13 Mb region examined (Fig. 3). LOH for chromosome 8 wasalso more likely to be complete LOH (Fig. 2), only a single caseof lung metastases showing partial LOH. The most frequentlyinvolved locus was D8S1786 potentially lost in all tumoursshowing LOH for chromosome 8. Where loss occurred, neitherregion showed preferential loss for either the maternal orpaternal contribution to the genome.

Deletion mapping in GTT following pregnancies with CHM

For all fourteen tumours following CHM, a single allelewas observed following amplification with markers forcontrol loci, chromosome 7q11.2 or chromosome 8p12–p21.Occasional results were uninformative in that the tumourDNA was homozygous for a marker present in both parentsbut, where informative, the single allele identified was alwayspaternal in origin (Fig. 2). Since all tumours were derived

Fig. 1. Representative electropherograms of LOH analysis showing microsatellite markers amplified from maternal, paternal and tumour DNA. (a) No LOH for controlmarker D7S519 and (b) allelic imbalance (partial LOH) for marker D7S669 in GTT originating in non-molar pregnancies. An allele from the patient and her partner(shaded) is present in both. However, in (b), the height of the maternally derived allele is less than that expected given the height of the paternally derived allele at thesame locus.

645B. Burke et al. / Gynecologic Oncology 103 (2006) 642–648

from homozygous CHM, LOH cannot be excluded. However,there was no evidence of homozygous deletions at any loci,DNA from all post-mole tumours being amenable to amplifi-cation with all markers. Despite careful microdissection and theuse of LCM, a small number of host (maternal) cells werepresent in the tissue fromwhich DNAwas extracted in several ofthe tumours. DNA from these cells was co-amplified with DNAfrom the tumour and easily identified (Fig. 2). The level ofcontaminating DNA in all cases of post-mole GTT remainedconstant following amplification with different markers, pro-viding further evidence that no homozygous deletions of tumourDNAwere present in the samples investigated.

Discussion

Although cytogenetic studies have not revealed specificchromosomal abnormalities associated with GTT, moleculargenetic studies have revealed both loss and gain of chromo-somal material in choriocarcinoma. Using microcell hybridsfollowed by microsatellite genotyping of choriocarcinoma celllines Matsuda et al. [10] demonstrated homozygous deletionsfor one or more markers in the 7p12–q11.23 region in seven ofeight choriocarcinoma cell lines, suggesting that deletion of thisregion might be a significant event in the development ofchoriocarcinoma. In support of this they were able to showsimilar homozygous deletions in DNA prepared from 10 of 14

post-mole tumours. Large homozygous deletions are relativelyrare, loss at the chromosomal level involving only one of thepair of chromosomes in most tumours. Tumour development isfavoured when this loss includes the normal allele of a tumoursuppressor gene that is already mutated on the otherchromosome. Such losses are usually identified as LOH whentumour DNA is compared with that of the host.

At least half of all GTT originate in HM [16] a figure that isprobably higher in reality, since many tumours arise not fromthe clinically antecedent pregnancy but from previous, oftenunrecognised, HM. Although both choriocarcinoma and PSTTcan originate from partial HM [17,18], it is generally CHM thatgive rise to GTT. These pregnancies usually arise by fertilisationof an anucleate egg by a single sperm that undergoesreduplication [19] and are hence homozygous for all of thenuclear genome. The observation of deletions for 7q11.2 in themajority of post-mole tumours [10] suggests that this might bean early event in the development of GTT. The observation thatthese deletions were homozygous led us to hypothesise that theymight already be present in the paternal genome prior tofertilisation in those CHM that would subsequently progress toGTT.

The findings of this study confirm the earlier observationthat there was no loss of markers centromeric to the 7q11.2region in choriocarcinoma or PSTT since marker D7S519 wasretained in all samples. Partial LOH of a second control marker

Fig. 2. Representative electropherograms of LOH analysis showing microsatellite markers amplified from maternal, paternal and tumour DNA. (a) Absence of apaternally derived allele for marker D7S520 and (b) absence of a maternally derived allele for marker D8S1786 in GTT originating in non-molar pregnancies. (c)shows a single paternally derived allele (shaded) found in an androgenetic tumour derived from a CHM. Arrows indicate additional small peaks representingcontamination of the tumour tissue by host (maternal) cells.

646 B. Burke et al. / Gynecologic Oncology 103 (2006) 642–648

D7S669, telomeric to the critical region, was observed for twocases. In one case this formed part of a larger region of lossencompassing the critical region. Significant LOH wasobserved for the 7q11.2 region in tumours following non-

Fig. 3. LOH analysis of chromosome 7q11.2 and 8p12–21 in choriocarcinoma and PSrepresent LOH, an asterisk indicating partial LOH; open circles represent no LOH;

molar pregnancies. However, the proportion of tumoursshowing LOH was less than that observed by Matsuda et al.[10]. This may reflect the fact that the tumours in the previousstudy included a number of cell lines that may have

TT. Only cases with at least one marker showing LOH are shown. Filled circlesgrey circles, uninformative markers.

647B. Burke et al. / Gynecologic Oncology 103 (2006) 642–648

accumulated further changes in culture or that the samplesinvestigated may have been derived from more advancedtumours. There was no evidence in the present study that LOHfor 7q11.2 was associated with more advanced tumours as LOHoccurred in three of the eight primary choriocarcinoma and onlyone of 11 choriocarcinoma in which metastatic tumour tissuewas investigated.

Although LOH for 7q11.2 was observed in tumoursfollowing non-molar pregnancies, the homozygous deletionspreviously identified in post-mole choriocarcinoma were notfound in the present study. DNA from all cases included in thestudy was successfully amplified with all markers in the region.It is unclear why the homozygous deletions previously reportedfor the critical region of chromosome 7 were not observed inthis study. In the present series archival material, rather thanDNA from frozen tissue, was used. However, we havepreviously demonstrated LOH for several markers in high-grade gliomas using the same techniques [20] and havedemonstrated LOH in tumours developing in non-molarpregnancies in this study. Failure to demonstrate homozygousdeletions is unlikely to be due to differences in the tumourgroups investigated since the present study included bothchoriocarcinoma and PSTT and both primary and metastatictumours.

Interestingly, the study by Ahmed et al. [11] failed to showsignificant LOH for chromosome 7 in a series of 12choriocarcinoma. Instead, they observed amplification of7q21–q31 and loss of 8p12–p21 as frequent events. Usingcomparative genome hybridisation Ahmed et al. [11] demon-strated deletion of 8p in five cases of choriocarcinoma includingloss of the 8p12–p22 region in the single case of choriocarci-noma that followed a pregnancy with CHM. In the presentstudy, significant LOH was observed for the 8p12–p21 regionin tumours following non-molar pregnancies confirming theprevious observation. Although LOH for this region cannot beexcluded in post-mole tumours, no homozygous deletions wereobserved. Again, all cases showed amplification with allmarkers in the 8p12–21 region.

In tumours following non-molar pregnancies LOH was seenfor both chromosome 7p11.2 and 8p21–21, four cases showingLOH for both regions. Loss was seen in both choriocarcinomaand PSTT. LOH for chromosome 7 often involved only a singlemarker, the most frequent loss being marker D7S520 althoughthis was not lost in all cases showing LOH. In contrast, mosttumours showing complete or partial LOH for chromosome8 showed loss of several markers. The minimal deleted regionencompasses D8S1786 and is flanked by D8S1731 and NEFL,a region shown to be involved in a number of malignancies andin which several tumour suppressor genes have already beenidentified [13]. Examination of further markers in the region isneeded to refine the specific region of loss and identify thosetumour suppressor genes that might be involved in thedevelopment of GTT.

Although significant LOH was found in tumours followingnon-molar pregnancies only 29% of tumours showed LOH foreach region suggesting that loss of tumour suppressor genesin these regions are late, rather than primary, events in the

malignant pathway and unlikely to be useful markers withwhich to identify those CHM at risk of progressing tomalignant disease. This is supported by the failure todemonstrate deletions of either region in post-mole GTT.Other markers are now needed to identify those CHM at riskof progressing to post-mole tumours. Recently, subtractivehybridisation has revealed a novel homeobox gene NECC1(not expressed in choriocarcinoma clone 1) [21], whiledifferential display has identified a number of novel genesexpressed in placenta [22]. Further studies will determinewhether any are potential prognostic markers in thedevelopment of GTT.

In addition to LOH and gene amplification, it is now wellrecognised that dysregulation of the normal methylationpatterns of imprinted genes may be associated with tumourdevelopment [23]. The relatively high propensity to malig-nancy of CHM could well be related to the abnormalexpression of imprinted genes that occurs as a consequenceof its androgenetic origin. In support of this women withfamilial recurrent hydatidiform mole, an autosomal recessivecondition in which affected individuals have an inheritedpredisposition to CHM, also have a high incidence of PTD.Despite the fact that HM in this condition are chromosom-ally normal, with a contribution from both parents, they havebeen shown to display the same abnormal patterns ofexpression and methylation status for imprinted genes as themore common androgenetic CHM [24]. Further investigationmay reveal that epigenetic factors play a more significantrole in the development of GTT than chromosomalabnormalities.

Acknowledgments

We are grateful to the patients and their partners who havetaken part in this study and to Sam Thornton and Ian Shore fortechnical assistance. This work was supported by grants fromWellBeing of Women and the Cancer Treatment and ResearchTrust.

References

[1] Bagshawe KD, Lawler SD. Choriocarcinoma. In: Schottenfeld D,Fraumeni JF, editors. Cancer Epidemiology and Prevention. Philadelphia:W.B. Saunders; 1982. p. 909–24.

[2] Tham BW, Everard JE, Tidy JA, Drew D, Hancock BW. Gestationaltrophoblastic disease in the Asian population of Northern England andNorth Wales. BJOG 2003;110:555–9.

[3] Li HW, Tsao SW, Cheung AN. Current understandings of the moleculargenetics of gestational trophoblastic diseases. Placenta 2002;23:20–31.

[4] Wake N, Tanaka K-i, Chapman V, Matsui S, Sandberg AA. Chromosomesand cellular origin of choriocarcinoma. Cancer Res 1981;41:3137–343.

[5] Sasaki S, Katayama PK, Roesler M, Pattillo RA, Mattingly RF, Ohkawa K.Cytogenetic analysis of choriocarcinoma cell lines. Acta Obstet GynaecolJpn 1982;34:2253–6.

[6] Sheppard DM, Fisher RA, Lawler SD. Karyotypic analysis andchromosome polymorphisms in four choriocarcinoma cell lines. CancerGenet Cytogenet 1985;16:251–9.

[7] Lawler S, Fisher RA. Genetic aspects of gestational trophoblastic tumours.In: Ichinoe K, editor. Trophoblastic Diseases. Tokyo: Igaku-Shoin; 1986.p. 23–33.

648 B. Burke et al. / Gynecologic Oncology 103 (2006) 642–648

[8] Bettio D, Giardino D, Rizzi N, Simoni G. Cytogenetic abnormalitiesdetected by direct analysis in a case of choriocarcinoma. Cancer GenetCytogenet 1993;68:149–51.

[9] Rodriguez E, Melamed J, Reuter V, Chaganti RS. Chromosomalabnormalities in choriocarcinomas of the female. Cancer Genet Cytogenet1995;80:9–12.

[10] Matsuda T, Sasaki M, Kato H, Yamada H, Cohen M, Barrett JC, et al.Human chromosome 7 carries a putative tumor suppressor gene(s)involved in choriocarcinoma. Oncogene 1997;15:2773–81.

[11] AhmedMN, Kim K, Haddad B, Berchuck A, Qumsiyeh MB. Comparativegenomic hybridization studies in hydatidiform moles and choriocarcino-ma: amplification of 7q21–q31 and loss of 8p12–p21 in choriocarcinoma.Cancer Genet Cytogenet 2004;116:10–5.

[12] Lassus H, Laitinen MP, Anttonen M, Heikinheimo M, Aaltonen LA,Ritvos O, et al. Comparison of serous and mucinous ovarian carcinomas:distinct pattern of allelic loss at distal 8p and expression of transcriptionfactor GATA-4. Lab Invest 2001;81:517–26.

[13] Armes JE, Hammet F, de Silva M, Ciciulla J, Ramus SJ, Soo WK, et al.Candidate tumor-suppressor genes on chromosome arm 8p in early-onsetand high-grade breast cancers. Oncogene 2004;23:5697–702.

[14] Kajii T, Ohama K. Androgenetic origin of hydatidiform mole. Nature1977;268:633–4.

[15] Fisher RA, Khatoon R, Paradinas FJ, Roberts AP, Newlands ES. Repetitivecomplete hydatidiform mole can be biparental in origin and either male orfemale. Hum Reprod 2000;15:594–8.

[16] WHO Scientific Group. Gestational Trophoblastic Disease. World HealthOrganisation Technical Report Series 1983;692.

[17] Seckl MJ, Fisher RA, Salerno G, Rees H, Paradinas FJ, Foskett M,et al. Choriocarcinoma and partial hydatidiform moles. Lancet 2000;356:36–9.

[18] Palmieri C, Fisher RA, Sebire NJ, Lindsay I, Smith JR, McCluggaggeWG,et al. Placental site trophoblastic tumour arising from a partial hydatidiformmole. Lancet 2005;366:688.

[19] Fisher RA, Povey S, Jeffreys AJ, Martin CA, Patel I, Lawler SD.Frequency of heterozygous complete hydatidiform moles, estimated bylocus-specific minisatellite and Y chromosome-specific probes. HumGenet 1989;82:259–63.

[20] Balesaria S, Brock C, Bower M, Clark J, Nicholson SK, Lewis P, et al.Loss of chromosome 10 is an independent prognostic factor in high gradegliomas. Br J Cancer 1999;81:1371–7.

[21] Asanoma K, Kato H, Inoue T, Matsuda T, Wake N. Analysis of a candidategene associated with growth suppression of choriocarcinoma anddifferentiation of trophoblasts. J Reprod Med 2004;49: 617–26.

[22] Garcia J, Castrillo JL. Differential display RT-PCR analysis of humanchoriocarcinoma cell lines and normal term trophoblast cells: identificationof new genes expressed in placenta. Placenta 2004;25:684–93.

[23] Feinberg AP, Tycko B. The history of cancer epigenetics. Nat Rev, Cancer2004;4:143–53.

[24] Fisher RA, Hodges MD, Newlands ES. Familial recurrent hydatidiformmole: a review. J Reprod Med 2004;49:595–601.