the origin of ovarian teratomas - journal of medical genetics

Journal of Medical Genetics, 1984, 21, 4-12

The origin of ovarian teratomas

JENNIFER M PARRINGTON, LYNNE F WEST, AND SUSAN POVEY

From the MRC Human Biochemical Genetics Unit, The Galton Laboratory, University College London,Wolfson House, 4 Stephenson Way, London NW] 2HE.

SUMMARY Chromosome and enzyme markers have been studied in 21 benign ovarian teratomasfrom 14 patients. Markers heterozygous in the patient were completely homozygous in 52 % of theteratomas and completely heterozygous in 19 %. The remainder showed a mixture of the two, 10 %having homozygous centromeres with some heterozygous enzyme markers and 19 % having hetero-zygous centromeres and some homozygous enzyme markers.These results suggest that benign ovarian teratomas in the present series arise from germ cells in a

number of different ways. Those with heterozygous centromeres probably arise by failure of meiosis 1.Some tumours with homozygous centromeres must arise by failure of meiosis 11, but because of thelow level of heterozygous enzyme markers in this group a substantial number are thought to arise byduplication of a mature ovum to give an entirely homozygous genotype, genetically the femaleequivalent of the complete hydatidiform mole.

Benign ovarian teratomas (dermoid cysts) arecommon tumours accounting for 10 to 20% of allovarian cysts. They are formed of fully differentiatedmature tissues, often from all three cell layers, andusually comprise a cyst containing sebaceousmaterial, hair, and often teeth.1 Most of them have anormal 46,XX karyotype2 but mosaicism has beenfound in some.3

Early studies showed that tissue extracts andcultured cells from teratomas were frequentlyhomozygous for enzyme markers which wereheterozygous in host cells.4 5 Moreover, all centro-meric chromosome markers heterozygous in thehost were subsequently shown to be homozygous inthe teratomas.6 This led to the conclusion thatthese tumours arose from a single germ cell after thefirst meiotic division. Studies in a strain of micewhere a high proportion develop spontaneousovarian teratomas indicate that these may arise inessentially the same way.8

If this interpretation is correct, analysis of geneticmarkers in teratomas should provide a means ofestimating the genetic distance between any givenlocus and the centromere, since heterozygositycould only arise as a result of recombination. Ofcourse, only teratomas occurring in subjects whohappen to be heterozygous for the locus in questionwould be informative. Assuming that all teratomasarise in the same way, Patil and co-workers3 calculatedReceived for publication 15 August 1983.Accepted for publication 20 August 1983.

4

the gene-to-centromere distances of phospho-glucomutase I (PGMI), phosphoglucomutase 3(PGM3), and phosphogluconate dehydrogenase(PGD) as 20, 17, and 33 centimorgans (cM), res-pectively. These figures contrast with estimates of68, 4, and 133 respectively which have been obtainedin females from family studies9 (table 1). From ourown data, we have previously observed that for lociknown to lie far from the centromere very fewtumours are heterozygous, leading to estimates ofgene-to-centromere distance much lower thanexpected.'1 We suggested that suppression of crossingover might occur in these tumours. However, analternative explanation would be that some ovarianteratomas arise by another mechanism.We have explored this possibility by detailed

investigation of ovarian teratomas in 14 patients.In each case karyotype analysis was performed onlymphocytes, on cells cultured from the teratoma,and where possible on cells cultured from normalsolid tissue. Enzyme analysis was carred out onthese cultures and also directly on the teratoma, onnormal solid tissue, and on blood.

Materials and methods

PATIENTS

Samples were obtained from 14 patients operated onat the National Temperance Hospital betweenDecember 1974 and August 1979. The patients wereunselected; all those in whom adequate samples and

copyright. on 26 D

ecember 2018 by guest. P

rotected byhttp://jm

g.bmj.com

/J M

ed Genet: first published as 10.1136/jm

g.21.1.4 on 1 February 1984. D

ownloaded from

http://jmg.bmj.com/


TABLE 1 Number of heterozygotes observed/total informative cases for 13 enzyme loci in 21 teratomas where thecentromere status is known compared with number of heterozygotes predictedfrom family data*.Enzyme Locus Chromosome Distance from Heterozygous Homozygous

arm centromere centromeres centromeres(Fc M) (presumed failure of (presumed failure of

meiosis I) meiosis II)

Observed Expected Observed Expected

Phosphogluconate dehydrogenase PGD 1 133 3/4 2-64 0/3 1.98x-fucosidase a-FUC llp 103 1/1 0.68 0/2 1.28Phosphoglucomutase I PGMI J 68 6/7 4.90 0/9 5.40Peptidase C PEPC lq 65 - - 0/1 0-59Acid phosphatase ACPI 2p 120 1/2 1.32 0/4 2-72Glyoxalase 1 GLO 6p 40 1/1 0.75 0/3 1-50Phosphoglucomutase 3 PGM3 6q 4 1/1 0.99 0/7 <0-07Adenylate kinase I AKI 9q 130 - - 0/1 0.68Phosphoglycolate phosphatase PGP 16p 60 - - 1/2 1-20Glutamate oxaloacetate GOT2 16q 34 1/1 0 -78 1/2 0-88transaminase (mitochondrial)

a-glucosidase a-GLU 17q 74 1/2 136 0/1 0 * 63Adenosine deaminase A DA 20q 80 1/1 0*65 0/3 2-10

Total 16/20 14 09 2/38 19-03Malic enzyme (mitochondrial) ME2 ? ? 2/4 ? 1/4 ?

*Predicted values calculated according to Sturt,1I using the map distances shown in column 4 and assuming at least one cross over per chromosomearm, no interference across the centromere, and random segregation of chromatids in meiosis II (fig 5, lb).

successful cultures were obtained are included.Ten patients had single cysts (five on the right,three on the left, two side not known) and fourpatients had bilateral cysts, three having one in eachovary and one patient (DC 131) having six tumourson the right and two on the left. All tumourscontained hair and sebaceous material and in mostcases a copy of the histological report on the tumourswas available, invariably confirming a diagnosis ofbenign ovarian teratoma. In two cases of multiplecysts (DC 86 and DC 131) one of the tumours was sosmall it was not sampled for fear of interfering withthe histological diagnosis. The age of the patientsranged from 20 to 49 years and in most cases theteratoma was an incidental finding, frequentlyduring pregnancy. None of the patients was knownto have taken any fertility drugs. One patient withbilateral cysts (DC 118, aged 30) ascertained duringpregnancy subsequently produced a child withtrisomy 21.

Material was collected directly from the operatingtheatre and always dissected by the same person(SP). A piece from the centre of the dermoid cystwas removed, taking care not to include any materialfrom the ovarian capsule. The part used for culturewas always taken from the growth nidus and nearlyalways contained active hair follicles. A specimen ofnormal tissue was also collected, either from theovary or Fallopian tube or, if this was not available,from the outer part of the cyst capsule. (Our previousstudies had shown that this is always geneticallyrepresentative of the patient rather than the cyst.)Small samples for culture were placed immediately inmedium, the rest of the sample being stored at

-20°C before enzyme analysis. The remainder ofthe specimen was sent for histology. Some days laterthe patient was visited on the ward and in all casesagreed to provide a blood sample after the projecthad been explained to her.

METHODSCell cultureSmall explants of the dermoid centre and of normaltissue were placed in Leighton tubes, immobilisedwith coverslips, and grown in Eagle's MEM supple-mented with 20% human serum or 10% fetal bovineserum (Flow Laboratories). Once fibroblast growthwas well established, all cultures were maintained inEagle's MEM containing 10% fetal bovine serum,Hepes buffer (10 mmol/l) (Hopkins and Williams),penicillin (100 IU/ml), and streptomycin (100 ,ug/ml).Chromosomepreparation andbandingChromosome preparations were normally madebetween the third and seventh passage in culture.Metaphases were collected for 1 to 4 hours withcolchicine, treated with hypotonic (0 075 mol/l KC1),fixed in acetic methanol, spread on cold slides, andair dried. Blood cultures were processed at 72 hoursin the same way.For Q banding, slides were stained in 0 5% atebrin

(GURR) in distilled water for 7 to 15 minutes,washed in tap water for 2 minutes, mounted in athin layer of 12% sucrose solution, and viewed with aZeiss fluorescence microscope.12 Q band hetero-morphisms were typed from photographs. At leastthree and up to 10 metaphases were examined foreach tissue and classified according to the intensityof fluorescence (fig 1).

5

copyright. on 26 D



g.bmj.com

/J M



ownloaded from

http://jmg.bmj.com/

J M Parrington, L F West, and S Povey

4.i

*0a

IIG 1 Q banided paritial kar votypes fron twod,iff6-ent patients showing heter-omorphisms onchromosomes 15 (DC 119) and 13 (DC 118). In bothpatients there is one homozygous teratoma. The leftteratoma from DC 118 is heterozygous like the host.(b-bright, N= normal).

A modification of the method of Sumner13 was

used for C banding. Slides were placed in 0.2 mol/lHCI for one hour, rinsed in deionised water, andtreated with 5% barium hydroxide at 50°C for 15seconds to 2 minutes depending on the age of theslide. They were then rinsed in deionised water,placed in 2 x SSC at 60°C for one hour, and stainedwith 10% Giemsa (BDH) at pH 6 8. C bands were

typed in at least 10 cells, both down the microscopeand from photographs. They were classifiedaccording to relative size and position in relation tothe centromere (fig 2).

Enzymtle anialysisTissues to be investigated directly were homo-genised in a Silverson emulsifier with an equalvolume of distilled water. After centrifugation at10 000 g for 10 minutes, the supernatant was usedfor enzyme analysis. Red cell haemolysates were

prepared by freezing and thawing packed washedred cells, and extracts of leucocytes and cultured cellswere prepared as previously described.14 15

Enzyme analysis was done mainly by horizontalstarch gel electrophoresis.16 17 The typing of o-

fucosidase and the subtyping of phosphoglucomutase(PGM I) were done by isoe'lectric focusing.18 19A list of the polymorphic enzyme loci studied,

together with what is known of their chromosomeassignment and distance from the centromere, isshown in table 1. The chromosomal assignment istaken from a recent Human Gene Mapping work-

FIG 2 C banded partial karyotvpes from two patientsshowing heteromorphisms on chromosomes I and 9(DC 125) and I (DC 131). The single teratom7a fromDC 125 is homnozygous thlroughout. Two of the seventeratomas from DC 131 are illustrated: R3(homozygous) and RI (heterozygous). (L = large,M_= medium, S= smnall).

shop.20 Map distances have been estimated fromfamily studies and are in female centimorgans.Many of the data are from Cook and co-workers,9female distances being obtained by doubling maledistances where no direct information was available,as suggested by these authors. More recent informa-tion was used for chromosome 121 and chromosome16.2 As the number of polymorphic markers avail-able increased during the span of this work, notevery sample was tested for every enzyme.

Results

CELL GROWTH

Good growth was obtained from most of the cystmaterial. The resulting fibroblast cultures had thegeneral appearance and growth pattern of those setup in our laboratory from skin biopsies. In contrast,good cultures from normal tissue were the exception.Growth after subculturing was always slow and itwas often difficult to obtain enough cells for chro-mosome and enzyme studies. Host typing wastherefore done mostly on blood, but where fibroblastcultures were obtained they were used asconfirmation.

KARYOTYPE

Nearly all teratomas had a normal female karyotypealthough a few abnormal cells were found in some

6

.-f* 0 14 A

4

copyright. on 26 D



g.bmj.com

/J M



ownloaded from

http://jmg.bmj.com/


o O 0 " 0 0 0 -

ri4 r 0 0 en~ en en r4 Ci

~c0 °0 't 0 0 0 0o

" 0ro n n en c --

(II ri4 ri4

" r-

clNrrie enCCCr

c- CA A I_ z

m

ri4 ri ri ri

4 It

IAN "4 _ , 'l + +

es4 C1

rI

Z Z

z Z Z. Z.c D0 D0 D D Z

- ; 2 23~a " Q4

C C C C c c

00) - -

*CO-uC* -

*0-

>0 )0 *C- 0)*0 * -

CZ 4,

p~X pC)p7 o p o Q o Cp- Q V fs, Q :'- a) m' 0

7

2:-C

IiEc1

20

0

0

00

0)C

01

0

0 -

E-_

>C0' 0

CLC6 CZ

o

0) .°2 m

- .00

0'.0=

* 110aQ

I

Il

I

copyright. on 26 D



g.bmj.com

/J M



ownloaded from

http://jmg.bmj.com/

J M Parrington, L F West, and S Povey

cultures. These are not thought to be relevant to thepresent discussion but will be the subject of a furthercommunication.

CHROMOSOME HETEROMORPHISMS

Clear cut differences between the centromericregions of at least two pairs of chromosomes were

identified in normal tissue from each patient.Where results were obtained from both blood andfibroblasts derived from normal host tissue they were

always in agreement. Chromosomes were studied in atotal of 21 teratomas (tables 2 and 3). In any one

tumour the centromere markers were either allheterozygous, as in the host (eight cases), or allhomozygous (13 cases). However, both types of cystcould occur in the same patient. Examples of bothhomozygous and heterozygous teratomas are shownin figs 1 and 2.

ENZYME RESULTS

The results of testing the phenotype of the patientwere consistent in blood and in other normal tissuestested, whether these were examined in freshpreparations or in culture. Some difficulty was

experienced in assessing PGM3 and ACPI in freshteratoma tissue. With the exception of DC 119 Rcyst, which failed to grow in culture but producedvery clear results from the tumour extract, all theenzyme results on teratomas described here were

confirmed in cell cultures.Tables 2 and 3 show the enzyme results for all

enzyme loci where the patient was heterozygous.In every case where the patient was homozygousthe cyst was also homozygous, and these uninform-ative data are not shown. Examples of differentenzyme phenotypes in multiple tumours from twodifferent patients are shown in figs 3 and 4.

In most of the teratomas the enzyme markers were

either all homozygous or all heterozygous, resembling

PGD phenotype

FIG 3 PGD isozymnes in cells culturedfroni seven

separate teroprmas occurring in DC 131.

:e

L0.'-0

o

Uz -t)2

C-0lb_ -

E~CJ

tA

q

zoi

-o5 Cl : o o5 -

N o eoooCOo '~t°4 OO m Ot

c) c)° °°

cl CA _ --

7 7

"-

Cl _ Cl

Cl Cl Cl

_ _

77N

7

U U UU

<uuu«<-

cnv) (A -4.4(A(A(

Z z£ Z

~ u

W 4-jo : .U0 4C 40

o

8

4.

- - - - -- -

_

7CI ,l Ir f

00

copyright. on 26 D



g.bmj.com

/J M



ownloaded from

http://jmg.bmj.com/

The origini of ovariant teratomlas

a GLU phenotype

FIG 4 Acid oe-glucosidose isozymes in a placentalcontrol and in cultured cells showing differenthomiozygous phenotypes in the two cYsts ofDC 118.Chromosomiie anal vsis on these cvsts (fig 1) showedone to have heterozYgous centromeres.

the results found for centromeric heteromorphisms.Four cysts were completely heterozygous for theenzyme and chromosome markers tested and thuswere indistinguishable from host tissue. Four cystsin the multiple group, while heterozygous at thecentromere, were homozygous for one or moreinformative enzyme markers. Only two cysts withhomozygous centromeres, one in the single and onein the multiple group, were found to have bothheterozygous and homozygous enzyme markers.Thus, taking all the teratomas together, 33 % of theinformative enzyme markers were heterozygous, butif we consider only the data from cysts with homo-zygous centromeres, the figure is much lower (7 %)(table 4).

In summary, therefore, in this small series of 21teratomas, 520% were completely homozygous, 190%completely heterozygous, 19% had heterozygouscentromeres with some homozygosity of enzyme

TABLE 4 EnzYme nsiarkers in 21 teratomiias arising in13 patients.

Diagnosis No of En,yme miarkers in cystsc)sis -_______-______

Homozygous Heterozygolos

Homozygous centromneresSingle cyst 8 22Multiple cysts 5 17

Heterozygous centromneresSingle cyst 2 0 6Multiple cysts 6 6 12

Total 21 45 21

markers, and 10% had homozygous centromereswith some heterozygous enzyme markers.

Discussion

Chromosome and enzyme data from ovarianteratomas in the present series have in most casesshown clear cut differences between 'host' andteratoma, supporting previous conclusions that thesetumours are of germ cell origin. In the four caseswhere such differences were not found, origin fromsomatic cells or, as in the earliest theories, from ablastomere which was an identical twin of the patientcannot be formally excluded. From the growthpattern of the cultures we feel fairly confident thatthe cells were indeed from the teratoma and notfrom contaminating host tissue.The teratomas studied here differ substantially

from those reported previously in that they showevidence of heterogeneous origin. In considering thepossible ways in which these tumours might ariseit is perhaps helpful to summarise the features ofnormal female meiosis, which involves two successivecell divisions referred to as meiosis I and 11. Meiosis Iis initiated in fetal life. Homologous chromosomesof primary oocytes pair and exchange geneticmaterial during the pachytene stage of prophase.Chromosomes remain as bivalents (dictyotene)throughout childhood and only progress to meta-phase I and cytokinesis in the adult ovary. Homo-logous chromosomes then separate into daughtercells, one of which is eliminated as the first polarbody. The chromosomes proceed directly intometaphase II without DNA synthesis and at thisstage ovulation occurs. The second maturationdivision (meiosis I1) is normally only completed iffertilisation occurs and takes place in the tuberather than the ovary. It involves separation of thetwo sister chromatids to give two haploid daughtercells, one of which is eliminated as the second polarbody. It appears that in the formation of ovarianteratomas, cells attempt to undergo, and sometimescomplete, both divisions within the ovary. Abnor-malities at various points in this process could resultin cells from which a diploid tumour might arise.We can consider them in developmental sequence inrelation to our own data.

FAILURE OF MEIOSIS I (FIG 5 1)Tumours arising by this mechanism would beheterozygous for chromosome and enzyme markersnear the centromere. Some homozygosity at moredistal loci would be expected to arise as a result ofcrossing over and segregation of homologous loci.All eight tumours heterozygous at the centromerescould have arisen in this way. The number of

9

Al: "'.EL I

copyright. on 26 D



g.bmj.com

/J M



ownloaded from

http://jmg.bmj.com/

10 J M Parrington, L F West, and S Povey

heterozygous enzyme loci in these tumours is showii from the centromere would be expected to bein table 1 and agrees closely with expected values. heterozygous in some cases as a result of crossing

over. This is the mode of origin which has beenFAILURE OF MEIOSIS II (FIG 5 Ii) suggested previously for ovarian teratomas,6 andTumours arising by this mechanism would be each of the 13 tumours homozygous at the centro-homozygous for centromere markers and for those mere in our own series could have arisen in this way.loci very close to the centromere. Loci more distal There is, however, a difficulty when all 13 tumours

meiosis I meiosis IIFormation of Formation of

Mechanism Primary oocytes 1st polar body 2nd polar body Ter3torma

A m AAMa suppressed X

I ~~~~I

b < suppressed or or

C MS--AM'

A SA S

A M A M

A = <) A Ma

C S ~~tC5- *suppressed c S

A M _43 (g3 C .... S A

b

C S~~~~~~ -#~~.suppressed s

C S ulcto

A ZcA

zzzzz~zz~ -.. C SC S duplication

A M A c sb C MIV ----r-----J

c S!- ~C SC s duplication

FIG 5 Possible mechanisms oforigin ofovarian teratomas. I. Suppression ofmeiosis 1. 11. Suppression ofmeiosis 1I. lII.Normal meiosis followed by duplication of haploid gamete. Each horizontal line represents a chromatid. a and b referto non-cross over and cross over respectively. A and C represent alleles for an enzyme marker some distance from thecentromere which could be distal to a cross over. M and S refer to medium and small C band centromere markersrespectively. The consequences of each mechanism for the teratoma are as follows: lIIa, Illb, and hIa. Both centromereand enzyme markers have to be homozygous. lIb. Centromeres homozygous, enzyme marker heterozygous. Ia. Bothcentromere and enzyme marker heterozygous. Ib. Centromeres always heterozygous, enzyme markers can be eitherhomozygous or heterozygous. *Chromosomes enclosed within an elipse denote those going into a polar body. There isan equal chance that either product of division will enter the polar body. Not all the possible combinations areillustrated.

copyright. on 26 D



g.bmj.com

/J M



ownloaded from

http://jmg.bmj.com/


are considered together, because the proportion ofmarkers in these tumours which are heterozygouseven at distal loci is extremely low (7 %).

Several authors have suggested functions whichwould relate heterozygosity (equivalent to seconddivision segregation in normal meiosis) to distancefrom the centromere in this situation.1023 Fortu-nately, all models agree that a very low level ofheterozygosity (less than 20%) would only beexpected within a short distance (<20 cM) from thecentromere. From table 1 it is clear that the onlylocus falling within this region is PGM3, and here ahigh proportion of homozygous teratomas would beexpected and was indeed observed.However, at the other loci there are far fewer

heterozygotes than expected, there being onlythree examples of heterozygosity out of 35 chancesamong tumours known to be homozygous forcentromere markers. One explanation for this wouldbe a reduced frequency of crossing over in meiosisleading to ovarian teratomas. This seems inherentlyrather unlikely since the amount of crossing over isdetermined during fetal life, so that it would benecessary to postulate an abnormality in some of theprimary oocytes at a very early stage. We have alsorecently obtained direct evidence that crossing overdoes occur in oocytes destined to become teratomasfrom detailed studies on one of the cases describedhere (DC 131).24 The results confirmed the existenceof crossing over in the short arm of chromosome 1 inall seven teratomas before development.A low level of heterozygosity together with clear

evidence of crossing over therefore suggest to us thata substantial number of teratomas do not arise byfailure of meiosis II. This contrasts with the situationin the inbred mouse line LT/Sv where elegantexperiments using teratomas in recombinant inbredstrains have given gene-to-gene map distancescomparable with those obtained by classical back-cross analysis.25 This provides strong evidence thatin the mouse these tumours can arise aftersuppression of meiosis II. The gene-to-centromeredistances estimated from the teratomas were ratherless than those obtained using Robertsonian trans-locations and pedigree analysis. Although thiscould possibly represent a small admixture of tera-tomas of different origin, the authors' suggestionthat Robertsonian translocations may change thefrequency of crossing over seems more probable.

ATTEMPTED DIVISION OF A MATURE OVUM(FIG 5 iii)If some stimulus caused an ovum to reach maturitywithin the ovary and then to attempt to divide butthe first cleavage was not successful, this wouldresult in a doubling up of the haploid chromosome

number. All chromosome and enzyme markerswould then be homozygous. The results on 11 of the13 teratomas homozygous at the centromere wouldbe compatible with this mode of origin.FUSION OF TWO INDEPENDENTLYDERIVED MATURE OVAIn a heterozygous patient there would be an equalchance of homozygosity or heterozygosity at eachlocus. Some centromere markers might be hetero-zygous and others homozygous in the same tumour.This was never found in our series, but this mode oforigin cannot be formally excluded in any particularcase.

FERTILISATIONThis would produce tumours containing genes andchromosome markers not present in the patient.This was not seen in any case.The most likely interpretation of our results seems

to be that the development of a teratoma is associatedwith an attempt to complete the whole of meiosis,and indeed to continue into development of azygote, within the ovary and without the stimulus offertilisation. In this abnormal environment there isa failure of the products of cell division to separate,and this can occur at several different stages givingtumours of different genetic constitution. The triggerfor these precocious divisions is quite unknown.From a practical point of view, the heterogeneousnature of these ovarian teratomas makes themcompletely useless for human gene mapping. Even ifextensive testing established the mode of origin in aparticular case, considerable bias would be intro-duced in this way. However, the results do reawakenquestions about the biology of germ cells and theirtumours.

It is of interest that, even with hindsight, we candetect no morphological differences between tumoursthought to arise in these different ways. It has beensuggested that the benign nature of teratomas maybe a reflection of their heterozygous genotype.26However, many of the benign tumours describedhere are apparently completely homozygous. Incontrast, malignant testicular teratomas showheterozygosity with respect to enzyme markers,centromeres, and the possession of a Y chromo-some27(and our unpublished observations).We cannot, of course, make any deduction from

the birth of an infant with trisomy 21 to one of thepatients with bilateral dermoid cysts. We could not,however, exclude any association between thesedifferent meiotic errors, and are not aware of anypublished data on this subject.

If our interpretation of the origin of ovarianteratomas is correct, it is curious that duplication of a

I1I

copyright. on 26 D



g.bmj.com

/J M



ownloaded from

http://jmg.bmj.com/

JM Parrington, L F West, and S Povey

haploid genome arising maternally in the ovaryproduces a dermoid cyst, whereas duplication of asperm genome growing in the uterus gives anotherbenign tumour, the hydatidiform mole.2829 Itwould be of interest to examine the origin of the raremalignant ovarian teratomas, as well as other germcell tumours, from the same point of view.

We are most grateful to Professor Fairweather andhis colleagues, all the theatre staff at the NationalTemperance hospital, and Dr P Hughesdon (patho-logist) for their patience and co-operation over manyyears. The expert technical assistance of GloriaJones, Steve Jeremiah, and Nona Parry-Jones werealso appreciated.

References

Fox H, Langley FA. Tumiours ojf the ovary. London:Heinemann, 1976.

2 Corfman PA, Richart RM. Chromosome number andmorphology of benign ovarian cystic teratomas. N Engl JMed 1964;271:1241-4.Patil SR, Kaiser-McCaw B, Hecht F, Linder D,Lourien EW. Human benign ovarian teratomas: chro-mosomal and electrophoretic enzyme studies. BirthDefects 1978;XIV:297-301.Linder D. Gene loss in human teratomas. Proc Natl AcalSci USA 1969;63:699-704.

5 Linder D, Power J. Further evidence for post-meioticorigin of teratomas in the human female. Ann Hum Genet1970;34:21-30.

6 Linder D, Kaiser-McCaw B, Hecht F. Parthenogenicorigin of benign ovarian teratomas. N Engl J Afedl 1975;292:63-6.

7 Kaiser-McCaw B, Hecht F, Linder D, et al. Ovarianteratomas: cytologic data. Cytogenet Cell Genet 1976;16:391-5.

8 Eppig JT, Kozak LP, Eicher EM, Stevens LC. Ovarianteratomas in. mice are derived from oocytes that havecompleted the first meiotic division. Natuire 1977;269:517-8.

9 Cook PJL, Noades JE, Lomas CG, Buckton- KE,Robson EB. Exclusion mapping illustrated by the MNSsblood group. Ann Hiln Genet 1980;44:61-73.

1 Sturt E. A mapping function for human chromosomes.Ann Hum Genet 1976;40:147-63.Parrington JM, West LF, Povey S. Gene mapping fromovarian teratomas. Cvtogenet Cell Genet 1979;25:196A.

12 Ganner E, Evans HJ. The relationship between patternsof DNA replication and of quinacrine fluorescence in thehuman chromosome complement. Chromwosomta 1971;35:326-41.

1 Sumner AT. A simple technique for demonstrating

centromeric heterochromatin. Exp Cell Res 1972;75:304-6.

14 King J, Cook PJL. Glucose dehydrogenase polymorphismin man. Ann Hum Genet 1981 ;45:129-34.

15 Kielty CM, Povey S, Hopkinson DA. Regulation ofexpression of liver-specific enzymes. 1. Detection inmammalian tissues and cultured cells. Ann Huim Genet1981 ;45:341-56.

16 Harris H, Hopkinson DA. Handbook of enzyme electro-phoresis in humnan genetics. Amsterdam: North-Holland,1976.

17 Harris H, Hopkinson DA. Handlbook oj enzyme electro-phoresis in human genetics. Supplement. Amsterdam:North-Holland, 1977.

18 Turner BM, Turner VS, Beratis NG, Hirschhorn K.Polymorphism of human alpha-fucosidase. Amn J Hum,Genet 1975;27:651-61.

19 Sutton JG, Burgess R. Genetic evidence for four commonialleles at the phosphoglucomutase-l locus (PGM 1)detectable by isoelectric focussing. Vax Sang 1978;34:97-103.

20 Human Gene Mapping 6 (1981). Sixth InternatioiialWorkshop on Human Gene Mapping. Cltogenet CellGenet 1982 ;32 :No 1-4.

21 King J. Linkage studies on human chromosome 1. PhDthesis, University of London, 1982.

22 Jeremiah SJ, Povey S, Cooke PJL, et al. Mapping studieson human mitochondrial glutamate oxaloacetate trans-aminase. Ann Hum Genet 1982;46:145-52.

23 Ott J, Linder D, Kaiser-McCaw B, Lovrien EW, Hecht F.Estimating distances from the centromere by means ofbenign ovarian teratomas in man. Ann Humin Genet1976;40:191-6.

24 Carritt B, Parrington JM, Welch HM, Povey S. Diverseorigins of multiple ovarian teratomas in a single individual.Proc Natl Acad Sci USA 1982 ;79 :7400-4.

2- Eppig JT, Eicher EM. Application of the ovarian tera-toma mapping method in the mouse. Genetics 1983;103:797-812.

26 Riley PA, Sutton PM. Why are ovarian teratomas benignwhilst teratomas of the testis are malignant? Lancet1975;i:1360-2.

27 Wang N, Perkins KL, Bronson DL, Fraley EE. Cyto-genetic evidence for premeiotic transformation ofhuman testicular cancers. Cancer Res 1981 ;41 :2135-40.

28 Lawler SD, Pickthall VJ, Fisher RA, Povey S, Evans MW,Szulman AE. Genetic studies of complete and partialhydatidiform moles. Lancet 1979;ii:580.

29 Jacobs PA, Wilson CM, Sprenkle JA, Rosenshein NB,Migeon BR. Mechanism of origin of hydatidiform moles.Nature 1980;286 :714-6.

Correspondence and requests for reprints toDr J M Parrington, MRC Human BiochemicalGenetics Unit, The Galton Laboratory, UniversityCollege London, Wolfson House, 4 StephensonWay, London NW 1 2HE.

12

copyright. on 26 D



g.bmj.com

/J M



ownloaded from

http://jmg.bmj.com/

the origin of ovarian teratomas - journal of medical genetics

Documents