GYNECOLOCIC
ONCOLOGY
47, 137-142 (1992)
Molecular Genetic Changes in Human Epithelial Ovarian Malignancies’ H. H. GALLION, M.D.,* D. E. POWELL, M.D.,? J. K. MORROW, PH.D.,t M. PIERETTI, PH.D.,*‘t E. CASE, M.S.,*,t M. S. TURKER, PH.D.,? P. D. DEPRIEST, M.D.,* J. E. HUNTER, M.D.,* AND J. R. VAN NAGELL, JR., M.D.* *Department
of Obstetrics 40536;
and Gynecology, and tDepartment
Division of Gynecologic of Pathology, University
Oncology, of Kentucky
University of Kentucky Medical Center, Lexington, Medical Center, Lexington, Kentucky 40536
Kentucky
Received Jantuary 22, 1992
The frequent finding of loss of heterozygosity (LOH) for a specificchromosomalmarker in tumor DNA comparedto normal DNA suggeststhe presenceof a closely linked tumor-suppressor gene.UsingSouthernblot analysis,34 primary ovarian epithelial tumors were examinedfor the presenceof tumor-specificallelic losses,usingsix probesfor chromosomes 6q, llp, 13q, 16q, and 17~. A high incidenceof LOH was observedon llp, 13q, and 17~. LOH for 17p waspresentin 3 of 4 (75%) informative benign ovarian tumors, 1 of 5 (20%) borderline tumors, and 16 of 24 (67%)invasive ovarian cancers.Allelic losswith the H-rasl probe on llp was presentin 10 of 19 (53%) invasive tumors but was not identified in 6 benignor borderlinetumors. LOH on 13qwas presentin 18 of 31 (58%) informative casesincluding 8 of 10 (80%) Stage 1 tumors. This preliminary study suggeststhat loss of tumor-suppressor geneson chromosomes 13qand 17p may be early eventsin ovarian tumorigenesisand that changeson chromosomelip are later events. 0 1992 Academic PRESS, I~C.
INTRODUCTION Recent evidence indicates that a normal cell is converted to its malignant couterpart following the accumulation of a critical number of mutations within regulatory genes [l]. These genes fall into two broad classes, protooncogenes which promote cell growth, and tumorsuppressor genes which inhibit cell growth. A mutation which activates one allele of a protooncogene can stimulate abnormal cell growth. Conversely, loss or inactivation of both alleles of a tumor-suppressor gene can remove normal constraints to cell proliferation. The essential role of tumor-suppressor genes in human tumors was established through the study of retinoblastoma, a Presented at the 23rd Annual Meeting of the Society of Gynecologic Oncologists, San Antonio, TX, March 15-18, 1992. ’ This research was supported by the following grants: The University of Kentucky Research Fund, GOG Basic Science Grant, and NIH Grant AG 08199.
rare childhood malignancy. Knudson hypothesized that familial retinoblastoma resulted from two independent mutations in the cellular genome, the first being a heritable germline mutation and the second a somatic mutation in the retinoblast [2]. Subsequent cytogenetic and molecular genetic analysis lead to the identification and cloning of the retinoblastoma tumor-suppressor gene and proved that the two mutational events proposed by Knudson were the inactivation of both alleles of this gene [35]. More recently, similar genetic analysis of human colon cancer has resulted in the identification of tumor-suppressor genes essential in colorectal tumorigenesis, the ~53 gene on chromosome 17, the DCC gene on chromosome 18, and the MCC gene on chromosome 5 [610]. Similarly, chromosomes lip, 13q, and 17p have been implicated as sites of putative tumor-suppressor genes in breast cancer [ll-131. Although ovarian cancer is the leading cause of gynecologic cancer death in the United States, little is known regarding the underlying molecular genetic changes which lead to the development of this biologically aggressive malignancy [ 141. Cytogenetic analysis of ovarian tumors by this group and others has demonstrated frequent structural aberrations of chromosomes 1, 3, 6, and 11, suggesting that inactivation of genes located on these chromosomes may be important in ovarian tumorigenesis [15-181. Recently, several investigators have reported a high frequency of allelic losses on chromosomes 3p, 6, llp, and 17 in ovarian tumors, suggesting that these chromosomes may contain tumor-suppressor genes important in ovarian tumorigenesis [19-231. This investigation was undertaken to (1) determine the frequency of allelic losses on chromosomes in the primary tumors of untreated patients with epithelial ovarian tumors and (2) correlate these losses with known prognostic indicators including FIG0 stage, grade, and cell type.
137 @NO-8258/92$4.00 Copyright 0 1992 by Academic Press,Inc. All rights of reproduction in any form reserved.
138
GALLION
MATERIAL
AND METHODS
From 1989 to 1991, blood and fresh primary tumor samples were obtained from 34 patients with previously untreated epithelial ovarian neoplasms, undergoing surgery at the Unviersity of Kentucky Medical Center. Tumor samples were obtained from the most cellular, least necrotic appearing portion of the fresh tumor, quick frozen and stored at -70°C. Tumors were classified histologically according to the WHO system and were graded as well, moderate, or poorly differentiated [24,25]. Tumor stage was assigned according to the 1985 FIG0 operative staging system [26]. Four of the tumors were benign. These included a germinal inclusion cyst, an adenofibroma, a Brenner tumor, and a serous cystadenoma. The most common malignant tumor was serous cystadenocarcinoma (15 patients), followed by endometrioid carcinoma (8 patients), mutinous cystadenocarcinoma (4 patients), and undifferentiated carcinoma (3 patients). Five tumors were borderline, 5 were grade 1, 11 were grade 2, and 9 were grade 3. Ten patients had FIG0 Stage I disease, 1 patient had Stage II disease, and 19 patients had Stage III disease. High molecular weight genomic DNA was extracted from stored tumor and blood by a nonorganic DNA extraction procedure [27]. Gel electrophoresis and Southern blotting was performed according to standard technique [28]. DNA from the patient’s blood was used as a control for the corresponding tumor DNA. A total of six loci from five different chromosomes were analyzed. Probes were obtained from American Type Culture Collection, (Rockville, MD). The following probes and restriction enzymes were used: pOR3 (Pvull), H-rasl (Mspl), H210 (EcoRl), hp2a (EcoRl), pYNZ22 (Mspl), and pUC1041 (Mspl). Cases were considered informative when heterozygosity was detected in the blood DNA preparation. Loss of heterozygosity (LOH) was defined as the absence of a hybridization band or a significant difference in intensity of a hybridization band in the tumor DNA specimen when compared to the control blood preparation from the same patient. RESULTS
Thirty-four primary epithelial ovarian tumors were analyzed using six probes on chromosomes 6q, lip, 13q, 16q, and 17~. Probes for chromosomes 6q, llp, and 17p were selected for analysis due to previous reports of karyotypic abnormalities and allelic losses of these chromosomes in epithelial ovarian carcinoma [15-221. The H-rasl probe was selected to characterize allelic losses on chromosome llp. Chromosome 13q, the site of the Rb gene, was analyzed as this gene appears to have general tumor suppressor properties in a number of cell types [12,29,30].
ET AL.
However, since most of our patients were not informative for the available Rb gene probe, the closely linked H210 probe was used. Chromosome 16 was selected as a control for LOH as there are no reports that this chromosome is involved in ovarian tumorigenesis. Thirty of the 34 tumors (88%) demonstrated LOH for at least one of the six probes used. The frequency of allelic losses at the specific loci examined are illustrated in Table 1. The incidence of LOH at the loci studied ranged from 8% with the hp2a probe on chromosome 16q to 58% with the H2-10 probe on chromosome 13q. Ten of the 25 (40%) informative cases for H-rasl demonstrated loss of heterozygosity. The Southern blot illustrated in Fig. 1 demonstrates loss of one allelic H-rasl restriction fragment in an ovarian tumor. Two distinct hybridization bands of 2.9- and 2.3-kb are present in the blood DNA lane. Therefore, the patient is an informative heterozygote. A marked difference in the intensity of the 2.3-kb band is seen in the tumor lane compared to the blood lane. This faint hybridization band is probably due to contamination of the tumor sample by stromal cells and is consistent with loss of heterozygosity at the H-rasl locus. Eighteen of the 31 (58?’ o ) cases informative for the H210 probe on chromosome 13q revealed LOH. Matched blood and tumor DNA samples hybridized with the H210 probe from 3 ovarian cancer patients are illustrated in Fig. 2. The first patient is noninformative for this probe as only one hybridization band is present in the blood lane. In the second case, two alleles are present in the blood DNA and loss of heterozygosity is observed in the matched tumor DNA. No loss of heterozygosity is present in the informative third case. Allelic loss for the pYNZ22 and pUClO-41 probe on chromosome 17p was present in 54 and 55% of the informative tumors, respectively. An ovarian tumor demonstrating LOH for both chromosome 17p probes is illustrated in Fig. 3. Allelic losses on chromosomes 6q and 16q were present at low frequencies (15 and 8%) respectively). The relationship between LOH and tumor grade is illustrated in Fig. 4. None of the six informative benign or borderline tumors exhibited LOH for llp, whereas allelic loss of llp was observed in 53% of invasive tumors. LOH on chromosome 13q was observed in one of four (25%) informative benign tumors, 40% of borderline tumors, and 70% of invasive tumors. Allelic losses for chromosome 17p was demonstrated in three of four (75%) informative benign tumors and 67% of invasive tumors. There was a direct relationship between increasing tumor grade and total number of LOH for probes on chromosomes lip, 13q, and 17~. The relationship between stage of disease and the frequency of LOH on chromosomes llp, 13q, and 17p is illustrated in Fig. 5. There was no significant difference
MOLECULAR
GENETICS
AND OVARIAN
139
CARCINOMA
TABLE 1 Frequency of Allelic Lossesat Specific Genetic Loci in Epithelial Ovarian Tumors
Locus ESR H-ras 1 D13S26 HP D17S5 D17S71
Probe
Chromosomal assignment
Enzyme
pOR3 pUCEJ66 H2-10 hp2o pYNZ22 pUClO-41
6q24-q27 llp15.5 13q21.1-q21.2 16q22.1 17p13.3 17p12-~11.2
Pvull Mspl EcoRl EcoRI MS@ Mspl
No. with LOH/ No. of informative cases
Loss (%)
2113 lo/25 18/31 2124 15/28 11120
15 40 58 08 54 55
Note. n = 34
in LOH for chromosome llp between Stage 1 and Stage II or III disease but the number of informative Stage I cases for lip is relatively low. A higher percentage of LOH for chromosome 13q was detected in Stage I tumors than that in more advanced stages of disease. A direct relationship between advanced stage and allelic loss on chromosome 17~ was observed. No significant difference in frequency of LOH was observed according to histologic cell type (data not shown). DISCUSSION Recent advances in molecular genetic techniques have made possible the detection of submicroscopic losses of genetic material in human tumors. The study of restriction fragment-length polymorphisms takes advantage of inherited differences in restriction enzyme sites between
maternal and paternal alleles. This results in different length DNA fragments that can be detected by Southern blot analysis. The detection of loss of heterozygosity in tumor DNA compared to normal DNA indicates loss or rearrangement of genetic material in the malignant cells. The finding of frequent allelic losses for a specific chromosome location suggests that a tumor-suppressor gene for the malignancy in question maps closely to the locus being sudied [ 11. Human colon cancer proved to be an ideal tumor type for genetic analysis due to the recognized progression of benign colon adenomas to frankly invasive tumors and because samples from colorectal tumors in various stages of development can be easily obtained for analysis. Although losses of genetic material in colorectal tumors were first detected cytogenetically, these abnormalities have been studied more extensively using molecular genetic probes [6,31]. For example, Vogelstein and co-workers observed LOH for chromosome 17~ in more than 75% of colorectal carcinomas but only rarely in adenomas (29,311. Within several individual tumors, 17~ allelic
- 2.9 kb
- 2.3 kb
probe: H-ras 1 (chromosome 1 lp) FIG. 1. Southern blot analysis demonstrating loss of one allelic Hrasl restriction fragment in ovarian cancer. DNA was isolated from the patient’s tumor lymphocytes and digested with Mspl. The sizes in kilobases of each fragment are given to the right of the panel. The faint 2.3-kb band in the tumor lane is probably due to partial contamination of the tumor sample with normal stromal cells.
- -15
kb
-
kb
-9
probe: H2-10 (chromosome 13q) FIG. 2. Representative Southern blot analysis of EcoRl digested DNA with probe H2-10 (13q21.1-q21.2) demonstrating (1) an uninformative homozygote, (2) allelic deletion in tumor DNA, and (3) no loss of heterozygosity in informative heterozygote (3).
140
GALLION
ET AL.
is consistent with the findings of other investigators [1923]. Of note, LOH at the estrogen receptor site on chromosome 6q was observed in only 15% of tumors. This is in contrast to the finding of both Lee et al. and Zheng et al. who reported allelic loss at the estrogen receptor locus in over 50% of ovarian tumors [19,22]. - 1.4kb Frequent allelic losses on chromosome 17p at the ~53 - 2.4 kb - 1.2 kb locus and other closely linked markers have been reported [19-211. In the series of poorly differentiated ovarian carcinomas reported by Lee et al., LOH on chromosome - 0.7 kb - 1.9 kb 17p was identified in 70% of cases [19]. Similarly, Eccles and co-workers reported LOH in 69% of ovarian tumors [20]. The high frequency of LOH observed in the current probe: pYNZ22 probe:pUClO-41 series is consistent with these reports. Although 17p allelic (chromosome 17~) (chromosome 17~) losses in benign ovarian tumors have not been previously reported, we observed LOH on chromosome 17p in three FIG. 3. Loss of heterozygosity on chromosomal segment 17~ using of four informative benign epithelial ovarian tumors but the indicated probles; DNA was digested with Mspl in both cases. borderline tumors. If this (A) Loss of one allele in the tumor sample is indicated by the missing in only one of five informative 1.4- and 1.2-kb bands, while the allele detected by the 0.7-kb band is difference is confirmed in larger numbers of benign and retained. (B) Probe pUClO-41 which has been previously mapped to borderline tumors, allele loss on 17p may be associated 17pll-qll, has now been more precisely assigned to the region 17p12~11.2. Loss of heterozygosity is detected in the tumor lane by the missing with progression from benign to frankly invasive tumors. This hypothesis would be consistent with recent histologic 2.4-kb band. data that indicates malignant ovarian tumors may arise directly from their benign counterparts [33,34]. losses were associated with progression from adenoma to We identified allelic loss with the H-rasl probe for carcinoma [32]. Similarly, reduction to homozygosity for chromosome llp in the majority of invasive tumors but a highly polymorphic marker on chromosome 18q was not in any of the benign or borderline tumors. This is found in the majority of advanced colon cancers [8]. consistent with the cytogenetic data previously reported from this institution in which abnormalities of chromoThese data suggested the presence of tumor-suppressor genes on these chromosomes and lead to the identification some 11 were common in invasive tumors but were not of the ~53 gene on chromosome 17p and a candidate identified in borderline tumors. Similarly, both Zheng and co-workers and Lee et al. noted LOH for llp in aptumor suppressor on chromosome 18, termed DCC [7,9]. proximately 50% of ovarian tumors [22,37]. These auMolecular changes in ovarian cancer have been more difficult to study than those in colon cancer for several thors noted that losses were most common in high grade reasons. These included the lack of a well-defined pro- or metastatic tumors, suggesting that llp loss may be gression from benign to malignant epithelium and the involved with later steps in ovarian tumorigenesis. Such relatively inaccessible location of the ovary within the LOH at the H-rasl locus on llp may be associated with peritoneal cavity. Moreover, the epithelial germinal in- a mutated H-ras allele on the remaining chromosome clusion cysts and benign cysts from which malignant ovar- which has been associated with the biologically aggressive ian tumors are thought to originate are usually composed behavior of high-grade tumors [35]. Alternatively, this of a single layer of epithelial cells, making it virtually may reflect a larger scale deletion on chromosome llp, impossible to obtain pure epithelium uncontaminated by resulting in loss of a yet unidentified tumor-suppressor adjacent stromal cells. Finally, karyotypic abnormalities gene on this chromosome. in ovarian tumors have been complex and difficult to Although LOH on chromosome 13q in advanced ovarinterpret. This may result from the advanced stage of the ian cancers has been previously reported by Li et al., it tumors studied. However, the consensus cytogenetic data was not observed as frequently as in the present study from this institution and others indicated that loci on [36]. In the current series, LOH for this chromosome was chromosomes 1, 3,6, and 11 could be involved in ovarian identified in the majority of Stage I tumors analyzed. This tumorigenesis [15-181. This cytogenetic data has served frequent finding of 13q allelic losses in both early and as a basis for ongoing molecular genetic studies. advanced-stage ovarian tumors suggests that loss of chroIn the current series, we determined the incidence of mosomal material from this chromosome may be a comtumor-specific allelic losses for probes on chromosomes mon event in ovarian tumorigenesis. Similar losses of 6, 11, 13, 16, and 17. A significant frequency of LOH chromosome 13q have been reported in human breast was observed for chromosomes lip, 13q, and 17~. This cancers [12]. Whether this reflects a deletion of the re-
MOLECULAR Chromosome
/
OK
AND OVARIAN
LMP
, Grade1
Chromosome
IV
/ Grade2
Grade3
/ Benign
17p
/
/ LMP
/ Grade
1
/ Grade
2
/ Grade
3
13q
/ LMP
/ Grade
1
/ Grade
2
/ Grade
3
Chromosomes lip. 13q. 17P
/
ov Benign
CARCINOMA Chromosome
11 p
,
Benign
GENETICS
/ Bemgn
/ LMP
/ Grade
1
I Grade
2
/ Grade
3
FIG. 4. Relationship between histologic grade and LOH in epithelial ovarian carcinomas. The first three panels are histograms representing the percentage of LOH for each tumor grade for the individual chromosomes studied (llp, 13q, 17~). The ratios above each histogram are the number of tumors with LOH over the total number of informative cases. The last panel summarizes the results for the three chromosomes. The percentages expressed by the histograms were calculated considering the total number of informative loci on chromosomes lip, 13q, and 17~.
q Stage
I
n Stage
II and III
0.6
0.4
0.2
tinoblastoma gene or of another tumor-suppressor gene on this chromosome is unknown. In summary, the findings of the present investigation indicate that allelic loss of chromosomes 13q, lip, and 17p are common in ovarian tumors. Although the number of benign or borderline tumors was small, our data suggests that loss of tumor-suppressor genes on chromosomes 13q and 17p may be early events in ovarian tumorigenesis and that changes on chromosome llp are later events. Future work will be directed toward using a larger number of molecular probes for each chromosome, both to confirm the present findings and to help localize the specific chromosome regions. Molecular genetic analysis of larger numbers of low grade and early stage tumors should provide further insight into ovarian tumorigenesis.
n llP
1%
17P
FIG. 5. Relationship between FIG0 stage of disease and the percentage of informative tumors with LOH. The histograms and ratios were calculated as described in the legend to Fig. 4.
ACKNOWLEDGMENT We give a special acknowledgment assistance.
to Monica L. Byrd for technical
142
GALLION
ET AL. IT
REFERENCES 1.
2. 3. 4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14. 15.
16.
17.
18.
Weinberg, R. A. Tumor suppressor genes, Science 254, 1138-1146 (1991). Knudson, A. G. Mutation and cancer: Statistical study of retinoblastoma, Proc. N&l. Acad. Sci. 68, 820-823 (1971). Yunis, J. J., and Ramsay, N. Retinoblastoma and subband deletion of chromosome 13, Am. J. Dis. Child. 132, 161-163 (1978). Benedict, W. F., Banerjec, A., Marck, C., and Murphree, A. L. Non-random chromosomal changes in direct preparations of primary retinoblastoma, Cancer Genet. Cytogenet. 10, 311-333 (1983). Lee, W. H., Bookstein, R., Hong, F., Young, L. J., Shew, J. Y., and Lee, E. Y. Human retinoblastoma susceptibility gene: Cloning, identification and sequence, Science, 1394-1399 (1987). Muleris, M., Salmon, R. J., Zafrani, B., Girodet, J., and Dutrillaux, B. Consistent deficiencies of chromosome 18 and of the short arm of chromosome 17 in eleven cases of human large bowel cancer: A possible recessive determinism, Ann. Genet. B(4), 206-213 (1985). Baker, S. S., Fearon, E. R., Nigro, J. M., Hamilton, S. R., Preisinger, A. C., Jessup, J. M., Vantuinen, P., Ledbetter, D. H., Barker, D. F., Nakamura, Y., White, R., and Vogelstein, B. Chromosome 17 deletions and p53 gene mutations in colorectal carcinomas, Science 244, 217 (1989). Vogelstein, B., Fearon, E. R., Hamilton, S. R., Preisinger, A. C., Leppert, M., Nakumura, Y., White, R., Smits, A. M. M., and Bos, J. L. Genetic alterations during colorectal-tumor development, N. Engl. J. Med. 319, 525-532 (1988). Fearon, E. R., Cho, K. R., Nigro, J. M., Kern, S. E., Simon, J. W., Ruppert, J. M., Hamilton, S. R., Preisinger, A. C., Thomas, G., Kinzler, K. W., and Vogelstein, B. Identification of a chromosome 18q gene that is altered in colorectal cancers, Science 247, 49-56 (1990). Kinzler, K. W., Nilbert, M. C., Vogelstein, B., Bryan, T. M., Levy, D. B., Smith, K. J., Preisinger, A. C., Hamilton, S. R., Hedge, P., Markham, A., Carison, M., Joslyn, G., Groden, J., White, R., Miki, Y., Miyoshi, Y., Nishisho, I., and Nakamura, Y. Identification of a gene located at chromosome 5q21 that is muted in colorectal cancers, Science 251, 1366-1370 (1991). Ah, I. U., Lidereau, R., Theillet, C., and Callahan, R. Reduction to homozygosity of genes on chromosome 11 in human breast neoplasia, Science 221, 185-188 (1987). Lundbert, C., Skoog, L., Cavenee, W. K., and Nordenskjold, M. Loss of heterozygosity in human ductal breast tumors indicates a recessive mutation on chromosome 13, Proc. Natl. Acad. Sci. 84, 2372-2376 (1987). Mackay, J., Steel, C. M., Elder, P. A., Forrest, A. P. M., and Evans, H. J. Allele less on short arm of chromosome 17 in breast cancer, Lancet 2, 1384-1385 (1988). Boring, C., Squires, T., and Tong, T. Cancer Statistics 1991, Cancer 41, 19-36 (1991). Atkin, N. B., and Baker, M. Abnormal chromosomes including small metacentrics in 14 ovarian cancers, Cancer Genet. Cytogenef. 26, 355-361 (1987). Tanaka, K., Boice, C. R., and Testa, J. R. Chromosome aberrations in nine patients with ovarian cancer, Cancer Genet. Cytogenet. 43, l-14 (1989). Gallion, H. H., Powell, D. E., Smith, L. W., Morrow, J. K., Martin, A. W., van Nagell, J. R., Jr., and Donaldson, E. S. Chromosome abnormalities in human epithelial ovarian malignancies, Gynecol. Oncol. 38, 473-477 (1990). Gallion, H. H., Powell, D. E., Smith, L. W., Vaugh, C. C., and
19.
20.
21.
22.
23.
24.
25.
26. 27.
28.
29.
Case, E. A. Cytogenetic changes in human epithelial ovarian cancer, Helen Harris Memorial Trust, in press. Lee, J. H., Kavanagh, J. J., Wildrick, D. M., Wharton, J. T., and Blick, M. Frequent loss of heterozygosity on chromosomes 6q, 11, and 17 in human ovarian carcinomas, Cancer Res. 50, 2724-2728 (1990). Eccles, D. M., Cranston, G., Steel, C. M., Nakamura, Y., and Leonard, R. C. F. Allele losses on chromosome 17 in human epithelial ovarian carcinoma, Oncogene 5, 1599-1601 (1990). Russell, S. E. H., Hickey, G. I., Lowry, W. S., White, P., and Atkinson, R. S. Allele loss from chromosome 17 in ovarian cancer, Oncogene 5, 1581-1583 (1990). Zheng, J., Robinson, W. R., Ehlen, T., Yu, M. C., and Dubeau, L. Distinction of low grade from high grade human ovarian carcinomas on the basis of losses of heterozygosity on chromosomes 3, 6, and 11 and HER-2/neu gene amplification, Cancer Res. 51, 4045-4051 (1991). Foulkes, W., Black, D., Solomon, E., and Trowsdale, J. Allele loss on chromosome 17q in sporadic ovarian cancer, Lancet 338, 444-445 (1991). Serov, S. F., Scully, R. E., and Sobin, L. H. International histologic classification of tumors. Histologic typing of ovarian tumors, World Health Organization (WHO) 9, 17-18 (1973). Hendrickson, M. R., and Kempson, R. L. Surgical pathology of the uterine corpus, in Major problems in pathology (J. L. Bennington, Ed.) Saunders, Philadelphia, p. 7 (1980). Pettersson, F. Annual report on the results of treatment in gynecological cancer, FIG0 20 (1987). Miller, S. A., Dykes, D. D., and Polesky, H. F. A simple salting out procedure for extracting DNA from human nucleated cells, Nucleic Acids Res. 16(3),1215 (1988). Molecular Cloning-A Laboratory Manual. T. Sambrook, E. F. Fritsch, and T. Maniatis, Eds.), Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, 2nd ed. (1989). Vogelstein, B., Fearon, E. R., Hamilton, S. R., Preisinger, A. C., Nakamura, Y., White, R., Smits, A. M. M., and Bos, J. L. Genetic alterations during colorectal-tumor development, N. Eng. J. Med.
319,525-532(1988). 30. Benedict, W. F., Fung, Y. T., and Murphree, A. L. The gene responsible for the development of retinoblastoma and osteosarcoma, Cancer 62, 1691-1694 (1988). 31. Fearon, E. R., and Vogelstein, B. A genetic model for colorectal tumorigenesis, Cell 61, 759-767 (1990). 32. Fearon, E. R., Hamilton, S. R., and Vogelstein, B. Clonal analysis of human colorectal tumors, Science 238, 193-197 (1987). 33. Stenback, F. Benign, borderline and malignant serous cystadenomas of the ovary, Pathol. Res. Pratt. 172,58-72(1981). 34. Puls, L. E., Powell, D. E., DePriest, P. D., Gallion, H. H., Hunter, J. E., Kryscio, R. J., and van Nagell, J. R., Jr. Transition from benign to malignant epithelium in mutinous and serous ovarian cystadenocarcinoma, Gynecol. Oncol., in press. 35. Atkin, N. B., and Baker, M. C. Deficiency of all or part of chromosome 11 in several types of cancer: Significance of a reduction in the number of normal chromosomes 11, Cytogenet. Cell Genet. 47, 106-107 (1988). 36. Li, S., Schwartz, P. E., Lee, W. H., Yang-Feng, T. L. Allele loss at the retinoblastoma locus in human ovarian cancer, J. Natl. Cancer Inst. 83(9), 637-640 (1991). 37. Lee, J. H., Kavanagh, J. J., Wharton, J. T., Wildrick, D. M., and Blick, M. Allele loss at the c-Ha-rasl locus in human ovarian cancer, Cancer Res. 49, 1220-1222 (1989).
ONCOLOGY
47, 137-142 (1992)
Molecular Genetic Changes in Human Epithelial Ovarian Malignancies’ H. H. GALLION, M.D.,* D. E. POWELL, M.D.,? J. K. MORROW, PH.D.,t M. PIERETTI, PH.D.,*‘t E. CASE, M.S.,*,t M. S. TURKER, PH.D.,? P. D. DEPRIEST, M.D.,* J. E. HUNTER, M.D.,* AND J. R. VAN NAGELL, JR., M.D.* *Department
of Obstetrics 40536;
and Gynecology, and tDepartment
Division of Gynecologic of Pathology, University
Oncology, of Kentucky
University of Kentucky Medical Center, Lexington, Medical Center, Lexington, Kentucky 40536
Kentucky
Received Jantuary 22, 1992
The frequent finding of loss of heterozygosity (LOH) for a specificchromosomalmarker in tumor DNA comparedto normal DNA suggeststhe presenceof a closely linked tumor-suppressor gene.UsingSouthernblot analysis,34 primary ovarian epithelial tumors were examinedfor the presenceof tumor-specificallelic losses,usingsix probesfor chromosomes 6q, llp, 13q, 16q, and 17~. A high incidenceof LOH was observedon llp, 13q, and 17~. LOH for 17p waspresentin 3 of 4 (75%) informative benign ovarian tumors, 1 of 5 (20%) borderline tumors, and 16 of 24 (67%)invasive ovarian cancers.Allelic losswith the H-rasl probe on llp was presentin 10 of 19 (53%) invasive tumors but was not identified in 6 benignor borderlinetumors. LOH on 13qwas presentin 18 of 31 (58%) informative casesincluding 8 of 10 (80%) Stage 1 tumors. This preliminary study suggeststhat loss of tumor-suppressor geneson chromosomes 13qand 17p may be early eventsin ovarian tumorigenesisand that changeson chromosomelip are later events. 0 1992 Academic PRESS, I~C.
INTRODUCTION Recent evidence indicates that a normal cell is converted to its malignant couterpart following the accumulation of a critical number of mutations within regulatory genes [l]. These genes fall into two broad classes, protooncogenes which promote cell growth, and tumorsuppressor genes which inhibit cell growth. A mutation which activates one allele of a protooncogene can stimulate abnormal cell growth. Conversely, loss or inactivation of both alleles of a tumor-suppressor gene can remove normal constraints to cell proliferation. The essential role of tumor-suppressor genes in human tumors was established through the study of retinoblastoma, a Presented at the 23rd Annual Meeting of the Society of Gynecologic Oncologists, San Antonio, TX, March 15-18, 1992. ’ This research was supported by the following grants: The University of Kentucky Research Fund, GOG Basic Science Grant, and NIH Grant AG 08199.
rare childhood malignancy. Knudson hypothesized that familial retinoblastoma resulted from two independent mutations in the cellular genome, the first being a heritable germline mutation and the second a somatic mutation in the retinoblast [2]. Subsequent cytogenetic and molecular genetic analysis lead to the identification and cloning of the retinoblastoma tumor-suppressor gene and proved that the two mutational events proposed by Knudson were the inactivation of both alleles of this gene [35]. More recently, similar genetic analysis of human colon cancer has resulted in the identification of tumor-suppressor genes essential in colorectal tumorigenesis, the ~53 gene on chromosome 17, the DCC gene on chromosome 18, and the MCC gene on chromosome 5 [610]. Similarly, chromosomes lip, 13q, and 17p have been implicated as sites of putative tumor-suppressor genes in breast cancer [ll-131. Although ovarian cancer is the leading cause of gynecologic cancer death in the United States, little is known regarding the underlying molecular genetic changes which lead to the development of this biologically aggressive malignancy [ 141. Cytogenetic analysis of ovarian tumors by this group and others has demonstrated frequent structural aberrations of chromosomes 1, 3, 6, and 11, suggesting that inactivation of genes located on these chromosomes may be important in ovarian tumorigenesis [15-181. Recently, several investigators have reported a high frequency of allelic losses on chromosomes 3p, 6, llp, and 17 in ovarian tumors, suggesting that these chromosomes may contain tumor-suppressor genes important in ovarian tumorigenesis [19-231. This investigation was undertaken to (1) determine the frequency of allelic losses on chromosomes in the primary tumors of untreated patients with epithelial ovarian tumors and (2) correlate these losses with known prognostic indicators including FIG0 stage, grade, and cell type.
137 @NO-8258/92$4.00 Copyright 0 1992 by Academic Press,Inc. All rights of reproduction in any form reserved.
138
GALLION
MATERIAL
AND METHODS
From 1989 to 1991, blood and fresh primary tumor samples were obtained from 34 patients with previously untreated epithelial ovarian neoplasms, undergoing surgery at the Unviersity of Kentucky Medical Center. Tumor samples were obtained from the most cellular, least necrotic appearing portion of the fresh tumor, quick frozen and stored at -70°C. Tumors were classified histologically according to the WHO system and were graded as well, moderate, or poorly differentiated [24,25]. Tumor stage was assigned according to the 1985 FIG0 operative staging system [26]. Four of the tumors were benign. These included a germinal inclusion cyst, an adenofibroma, a Brenner tumor, and a serous cystadenoma. The most common malignant tumor was serous cystadenocarcinoma (15 patients), followed by endometrioid carcinoma (8 patients), mutinous cystadenocarcinoma (4 patients), and undifferentiated carcinoma (3 patients). Five tumors were borderline, 5 were grade 1, 11 were grade 2, and 9 were grade 3. Ten patients had FIG0 Stage I disease, 1 patient had Stage II disease, and 19 patients had Stage III disease. High molecular weight genomic DNA was extracted from stored tumor and blood by a nonorganic DNA extraction procedure [27]. Gel electrophoresis and Southern blotting was performed according to standard technique [28]. DNA from the patient’s blood was used as a control for the corresponding tumor DNA. A total of six loci from five different chromosomes were analyzed. Probes were obtained from American Type Culture Collection, (Rockville, MD). The following probes and restriction enzymes were used: pOR3 (Pvull), H-rasl (Mspl), H210 (EcoRl), hp2a (EcoRl), pYNZ22 (Mspl), and pUC1041 (Mspl). Cases were considered informative when heterozygosity was detected in the blood DNA preparation. Loss of heterozygosity (LOH) was defined as the absence of a hybridization band or a significant difference in intensity of a hybridization band in the tumor DNA specimen when compared to the control blood preparation from the same patient. RESULTS
Thirty-four primary epithelial ovarian tumors were analyzed using six probes on chromosomes 6q, lip, 13q, 16q, and 17~. Probes for chromosomes 6q, llp, and 17p were selected for analysis due to previous reports of karyotypic abnormalities and allelic losses of these chromosomes in epithelial ovarian carcinoma [15-221. The H-rasl probe was selected to characterize allelic losses on chromosome llp. Chromosome 13q, the site of the Rb gene, was analyzed as this gene appears to have general tumor suppressor properties in a number of cell types [12,29,30].
ET AL.
However, since most of our patients were not informative for the available Rb gene probe, the closely linked H210 probe was used. Chromosome 16 was selected as a control for LOH as there are no reports that this chromosome is involved in ovarian tumorigenesis. Thirty of the 34 tumors (88%) demonstrated LOH for at least one of the six probes used. The frequency of allelic losses at the specific loci examined are illustrated in Table 1. The incidence of LOH at the loci studied ranged from 8% with the hp2a probe on chromosome 16q to 58% with the H2-10 probe on chromosome 13q. Ten of the 25 (40%) informative cases for H-rasl demonstrated loss of heterozygosity. The Southern blot illustrated in Fig. 1 demonstrates loss of one allelic H-rasl restriction fragment in an ovarian tumor. Two distinct hybridization bands of 2.9- and 2.3-kb are present in the blood DNA lane. Therefore, the patient is an informative heterozygote. A marked difference in the intensity of the 2.3-kb band is seen in the tumor lane compared to the blood lane. This faint hybridization band is probably due to contamination of the tumor sample by stromal cells and is consistent with loss of heterozygosity at the H-rasl locus. Eighteen of the 31 (58?’ o ) cases informative for the H210 probe on chromosome 13q revealed LOH. Matched blood and tumor DNA samples hybridized with the H210 probe from 3 ovarian cancer patients are illustrated in Fig. 2. The first patient is noninformative for this probe as only one hybridization band is present in the blood lane. In the second case, two alleles are present in the blood DNA and loss of heterozygosity is observed in the matched tumor DNA. No loss of heterozygosity is present in the informative third case. Allelic loss for the pYNZ22 and pUClO-41 probe on chromosome 17p was present in 54 and 55% of the informative tumors, respectively. An ovarian tumor demonstrating LOH for both chromosome 17p probes is illustrated in Fig. 3. Allelic losses on chromosomes 6q and 16q were present at low frequencies (15 and 8%) respectively). The relationship between LOH and tumor grade is illustrated in Fig. 4. None of the six informative benign or borderline tumors exhibited LOH for llp, whereas allelic loss of llp was observed in 53% of invasive tumors. LOH on chromosome 13q was observed in one of four (25%) informative benign tumors, 40% of borderline tumors, and 70% of invasive tumors. Allelic losses for chromosome 17p was demonstrated in three of four (75%) informative benign tumors and 67% of invasive tumors. There was a direct relationship between increasing tumor grade and total number of LOH for probes on chromosomes lip, 13q, and 17~. The relationship between stage of disease and the frequency of LOH on chromosomes llp, 13q, and 17p is illustrated in Fig. 5. There was no significant difference
MOLECULAR
GENETICS
AND OVARIAN
139
CARCINOMA
TABLE 1 Frequency of Allelic Lossesat Specific Genetic Loci in Epithelial Ovarian Tumors
Locus ESR H-ras 1 D13S26 HP D17S5 D17S71
Probe
Chromosomal assignment
Enzyme
pOR3 pUCEJ66 H2-10 hp2o pYNZ22 pUClO-41
6q24-q27 llp15.5 13q21.1-q21.2 16q22.1 17p13.3 17p12-~11.2
Pvull Mspl EcoRl EcoRI MS@ Mspl
No. with LOH/ No. of informative cases
Loss (%)
2113 lo/25 18/31 2124 15/28 11120
15 40 58 08 54 55
Note. n = 34
in LOH for chromosome llp between Stage 1 and Stage II or III disease but the number of informative Stage I cases for lip is relatively low. A higher percentage of LOH for chromosome 13q was detected in Stage I tumors than that in more advanced stages of disease. A direct relationship between advanced stage and allelic loss on chromosome 17~ was observed. No significant difference in frequency of LOH was observed according to histologic cell type (data not shown). DISCUSSION Recent advances in molecular genetic techniques have made possible the detection of submicroscopic losses of genetic material in human tumors. The study of restriction fragment-length polymorphisms takes advantage of inherited differences in restriction enzyme sites between
maternal and paternal alleles. This results in different length DNA fragments that can be detected by Southern blot analysis. The detection of loss of heterozygosity in tumor DNA compared to normal DNA indicates loss or rearrangement of genetic material in the malignant cells. The finding of frequent allelic losses for a specific chromosome location suggests that a tumor-suppressor gene for the malignancy in question maps closely to the locus being sudied [ 11. Human colon cancer proved to be an ideal tumor type for genetic analysis due to the recognized progression of benign colon adenomas to frankly invasive tumors and because samples from colorectal tumors in various stages of development can be easily obtained for analysis. Although losses of genetic material in colorectal tumors were first detected cytogenetically, these abnormalities have been studied more extensively using molecular genetic probes [6,31]. For example, Vogelstein and co-workers observed LOH for chromosome 17~ in more than 75% of colorectal carcinomas but only rarely in adenomas (29,311. Within several individual tumors, 17~ allelic
- 2.9 kb
- 2.3 kb
probe: H-ras 1 (chromosome 1 lp) FIG. 1. Southern blot analysis demonstrating loss of one allelic Hrasl restriction fragment in ovarian cancer. DNA was isolated from the patient’s tumor lymphocytes and digested with Mspl. The sizes in kilobases of each fragment are given to the right of the panel. The faint 2.3-kb band in the tumor lane is probably due to partial contamination of the tumor sample with normal stromal cells.
- -15
kb
-
kb
-9
probe: H2-10 (chromosome 13q) FIG. 2. Representative Southern blot analysis of EcoRl digested DNA with probe H2-10 (13q21.1-q21.2) demonstrating (1) an uninformative homozygote, (2) allelic deletion in tumor DNA, and (3) no loss of heterozygosity in informative heterozygote (3).
140
GALLION
ET AL.
is consistent with the findings of other investigators [1923]. Of note, LOH at the estrogen receptor site on chromosome 6q was observed in only 15% of tumors. This is in contrast to the finding of both Lee et al. and Zheng et al. who reported allelic loss at the estrogen receptor locus in over 50% of ovarian tumors [19,22]. - 1.4kb Frequent allelic losses on chromosome 17p at the ~53 - 2.4 kb - 1.2 kb locus and other closely linked markers have been reported [19-211. In the series of poorly differentiated ovarian carcinomas reported by Lee et al., LOH on chromosome - 0.7 kb - 1.9 kb 17p was identified in 70% of cases [19]. Similarly, Eccles and co-workers reported LOH in 69% of ovarian tumors [20]. The high frequency of LOH observed in the current probe: pYNZ22 probe:pUClO-41 series is consistent with these reports. Although 17p allelic (chromosome 17~) (chromosome 17~) losses in benign ovarian tumors have not been previously reported, we observed LOH on chromosome 17p in three FIG. 3. Loss of heterozygosity on chromosomal segment 17~ using of four informative benign epithelial ovarian tumors but the indicated probles; DNA was digested with Mspl in both cases. borderline tumors. If this (A) Loss of one allele in the tumor sample is indicated by the missing in only one of five informative 1.4- and 1.2-kb bands, while the allele detected by the 0.7-kb band is difference is confirmed in larger numbers of benign and retained. (B) Probe pUClO-41 which has been previously mapped to borderline tumors, allele loss on 17p may be associated 17pll-qll, has now been more precisely assigned to the region 17p12~11.2. Loss of heterozygosity is detected in the tumor lane by the missing with progression from benign to frankly invasive tumors. This hypothesis would be consistent with recent histologic 2.4-kb band. data that indicates malignant ovarian tumors may arise directly from their benign counterparts [33,34]. losses were associated with progression from adenoma to We identified allelic loss with the H-rasl probe for carcinoma [32]. Similarly, reduction to homozygosity for chromosome llp in the majority of invasive tumors but a highly polymorphic marker on chromosome 18q was not in any of the benign or borderline tumors. This is found in the majority of advanced colon cancers [8]. consistent with the cytogenetic data previously reported from this institution in which abnormalities of chromoThese data suggested the presence of tumor-suppressor genes on these chromosomes and lead to the identification some 11 were common in invasive tumors but were not of the ~53 gene on chromosome 17p and a candidate identified in borderline tumors. Similarly, both Zheng and co-workers and Lee et al. noted LOH for llp in aptumor suppressor on chromosome 18, termed DCC [7,9]. proximately 50% of ovarian tumors [22,37]. These auMolecular changes in ovarian cancer have been more difficult to study than those in colon cancer for several thors noted that losses were most common in high grade reasons. These included the lack of a well-defined pro- or metastatic tumors, suggesting that llp loss may be gression from benign to malignant epithelium and the involved with later steps in ovarian tumorigenesis. Such relatively inaccessible location of the ovary within the LOH at the H-rasl locus on llp may be associated with peritoneal cavity. Moreover, the epithelial germinal in- a mutated H-ras allele on the remaining chromosome clusion cysts and benign cysts from which malignant ovar- which has been associated with the biologically aggressive ian tumors are thought to originate are usually composed behavior of high-grade tumors [35]. Alternatively, this of a single layer of epithelial cells, making it virtually may reflect a larger scale deletion on chromosome llp, impossible to obtain pure epithelium uncontaminated by resulting in loss of a yet unidentified tumor-suppressor adjacent stromal cells. Finally, karyotypic abnormalities gene on this chromosome. in ovarian tumors have been complex and difficult to Although LOH on chromosome 13q in advanced ovarinterpret. This may result from the advanced stage of the ian cancers has been previously reported by Li et al., it tumors studied. However, the consensus cytogenetic data was not observed as frequently as in the present study from this institution and others indicated that loci on [36]. In the current series, LOH for this chromosome was chromosomes 1, 3,6, and 11 could be involved in ovarian identified in the majority of Stage I tumors analyzed. This tumorigenesis [15-181. This cytogenetic data has served frequent finding of 13q allelic losses in both early and as a basis for ongoing molecular genetic studies. advanced-stage ovarian tumors suggests that loss of chroIn the current series, we determined the incidence of mosomal material from this chromosome may be a comtumor-specific allelic losses for probes on chromosomes mon event in ovarian tumorigenesis. Similar losses of 6, 11, 13, 16, and 17. A significant frequency of LOH chromosome 13q have been reported in human breast was observed for chromosomes lip, 13q, and 17~. This cancers [12]. Whether this reflects a deletion of the re-
MOLECULAR Chromosome
/
OK
AND OVARIAN
LMP
, Grade1
Chromosome
IV
/ Grade2
Grade3
/ Benign
17p
/
/ LMP
/ Grade
1
/ Grade
2
/ Grade
3
13q
/ LMP
/ Grade
1
/ Grade
2
/ Grade
3
Chromosomes lip. 13q. 17P
/
ov Benign
CARCINOMA Chromosome
11 p
,
Benign
GENETICS
/ Bemgn
/ LMP
/ Grade
1
I Grade
2
/ Grade
3
FIG. 4. Relationship between histologic grade and LOH in epithelial ovarian carcinomas. The first three panels are histograms representing the percentage of LOH for each tumor grade for the individual chromosomes studied (llp, 13q, 17~). The ratios above each histogram are the number of tumors with LOH over the total number of informative cases. The last panel summarizes the results for the three chromosomes. The percentages expressed by the histograms were calculated considering the total number of informative loci on chromosomes lip, 13q, and 17~.
q Stage
I
n Stage
II and III
0.6
0.4
0.2
tinoblastoma gene or of another tumor-suppressor gene on this chromosome is unknown. In summary, the findings of the present investigation indicate that allelic loss of chromosomes 13q, lip, and 17p are common in ovarian tumors. Although the number of benign or borderline tumors was small, our data suggests that loss of tumor-suppressor genes on chromosomes 13q and 17p may be early events in ovarian tumorigenesis and that changes on chromosome llp are later events. Future work will be directed toward using a larger number of molecular probes for each chromosome, both to confirm the present findings and to help localize the specific chromosome regions. Molecular genetic analysis of larger numbers of low grade and early stage tumors should provide further insight into ovarian tumorigenesis.
n llP
1%
17P
FIG. 5. Relationship between FIG0 stage of disease and the percentage of informative tumors with LOH. The histograms and ratios were calculated as described in the legend to Fig. 4.
ACKNOWLEDGMENT We give a special acknowledgment assistance.
to Monica L. Byrd for technical
142
GALLION
ET AL. IT
REFERENCES 1.
2. 3. 4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14. 15.
16.
17.
18.
Weinberg, R. A. Tumor suppressor genes, Science 254, 1138-1146 (1991). Knudson, A. G. Mutation and cancer: Statistical study of retinoblastoma, Proc. N&l. Acad. Sci. 68, 820-823 (1971). Yunis, J. J., and Ramsay, N. Retinoblastoma and subband deletion of chromosome 13, Am. J. Dis. Child. 132, 161-163 (1978). Benedict, W. F., Banerjec, A., Marck, C., and Murphree, A. L. Non-random chromosomal changes in direct preparations of primary retinoblastoma, Cancer Genet. Cytogenet. 10, 311-333 (1983). Lee, W. H., Bookstein, R., Hong, F., Young, L. J., Shew, J. Y., and Lee, E. Y. Human retinoblastoma susceptibility gene: Cloning, identification and sequence, Science, 1394-1399 (1987). Muleris, M., Salmon, R. J., Zafrani, B., Girodet, J., and Dutrillaux, B. Consistent deficiencies of chromosome 18 and of the short arm of chromosome 17 in eleven cases of human large bowel cancer: A possible recessive determinism, Ann. Genet. B(4), 206-213 (1985). Baker, S. S., Fearon, E. R., Nigro, J. M., Hamilton, S. R., Preisinger, A. C., Jessup, J. M., Vantuinen, P., Ledbetter, D. H., Barker, D. F., Nakamura, Y., White, R., and Vogelstein, B. Chromosome 17 deletions and p53 gene mutations in colorectal carcinomas, Science 244, 217 (1989). Vogelstein, B., Fearon, E. R., Hamilton, S. R., Preisinger, A. C., Leppert, M., Nakumura, Y., White, R., Smits, A. M. M., and Bos, J. L. Genetic alterations during colorectal-tumor development, N. Engl. J. Med. 319, 525-532 (1988). Fearon, E. R., Cho, K. R., Nigro, J. M., Kern, S. E., Simon, J. W., Ruppert, J. M., Hamilton, S. R., Preisinger, A. C., Thomas, G., Kinzler, K. W., and Vogelstein, B. Identification of a chromosome 18q gene that is altered in colorectal cancers, Science 247, 49-56 (1990). Kinzler, K. W., Nilbert, M. C., Vogelstein, B., Bryan, T. M., Levy, D. B., Smith, K. J., Preisinger, A. C., Hamilton, S. R., Hedge, P., Markham, A., Carison, M., Joslyn, G., Groden, J., White, R., Miki, Y., Miyoshi, Y., Nishisho, I., and Nakamura, Y. Identification of a gene located at chromosome 5q21 that is muted in colorectal cancers, Science 251, 1366-1370 (1991). Ah, I. U., Lidereau, R., Theillet, C., and Callahan, R. Reduction to homozygosity of genes on chromosome 11 in human breast neoplasia, Science 221, 185-188 (1987). Lundbert, C., Skoog, L., Cavenee, W. K., and Nordenskjold, M. Loss of heterozygosity in human ductal breast tumors indicates a recessive mutation on chromosome 13, Proc. Natl. Acad. Sci. 84, 2372-2376 (1987). Mackay, J., Steel, C. M., Elder, P. A., Forrest, A. P. M., and Evans, H. J. Allele less on short arm of chromosome 17 in breast cancer, Lancet 2, 1384-1385 (1988). Boring, C., Squires, T., and Tong, T. Cancer Statistics 1991, Cancer 41, 19-36 (1991). Atkin, N. B., and Baker, M. Abnormal chromosomes including small metacentrics in 14 ovarian cancers, Cancer Genet. Cytogenef. 26, 355-361 (1987). Tanaka, K., Boice, C. R., and Testa, J. R. Chromosome aberrations in nine patients with ovarian cancer, Cancer Genet. Cytogenet. 43, l-14 (1989). Gallion, H. H., Powell, D. E., Smith, L. W., Morrow, J. K., Martin, A. W., van Nagell, J. R., Jr., and Donaldson, E. S. Chromosome abnormalities in human epithelial ovarian malignancies, Gynecol. Oncol. 38, 473-477 (1990). Gallion, H. H., Powell, D. E., Smith, L. W., Vaugh, C. C., and
19.
20.
21.
22.
23.
24.
25.
26. 27.
28.
29.
Case, E. A. Cytogenetic changes in human epithelial ovarian cancer, Helen Harris Memorial Trust, in press. Lee, J. H., Kavanagh, J. J., Wildrick, D. M., Wharton, J. T., and Blick, M. Frequent loss of heterozygosity on chromosomes 6q, 11, and 17 in human ovarian carcinomas, Cancer Res. 50, 2724-2728 (1990). Eccles, D. M., Cranston, G., Steel, C. M., Nakamura, Y., and Leonard, R. C. F. Allele losses on chromosome 17 in human epithelial ovarian carcinoma, Oncogene 5, 1599-1601 (1990). Russell, S. E. H., Hickey, G. I., Lowry, W. S., White, P., and Atkinson, R. S. Allele loss from chromosome 17 in ovarian cancer, Oncogene 5, 1581-1583 (1990). Zheng, J., Robinson, W. R., Ehlen, T., Yu, M. C., and Dubeau, L. Distinction of low grade from high grade human ovarian carcinomas on the basis of losses of heterozygosity on chromosomes 3, 6, and 11 and HER-2/neu gene amplification, Cancer Res. 51, 4045-4051 (1991). Foulkes, W., Black, D., Solomon, E., and Trowsdale, J. Allele loss on chromosome 17q in sporadic ovarian cancer, Lancet 338, 444-445 (1991). Serov, S. F., Scully, R. E., and Sobin, L. H. International histologic classification of tumors. Histologic typing of ovarian tumors, World Health Organization (WHO) 9, 17-18 (1973). Hendrickson, M. R., and Kempson, R. L. Surgical pathology of the uterine corpus, in Major problems in pathology (J. L. Bennington, Ed.) Saunders, Philadelphia, p. 7 (1980). Pettersson, F. Annual report on the results of treatment in gynecological cancer, FIG0 20 (1987). Miller, S. A., Dykes, D. D., and Polesky, H. F. A simple salting out procedure for extracting DNA from human nucleated cells, Nucleic Acids Res. 16(3),1215 (1988). Molecular Cloning-A Laboratory Manual. T. Sambrook, E. F. Fritsch, and T. Maniatis, Eds.), Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, 2nd ed. (1989). Vogelstein, B., Fearon, E. R., Hamilton, S. R., Preisinger, A. C., Nakamura, Y., White, R., Smits, A. M. M., and Bos, J. L. Genetic alterations during colorectal-tumor development, N. Eng. J. Med.
319,525-532(1988). 30. Benedict, W. F., Fung, Y. T., and Murphree, A. L. The gene responsible for the development of retinoblastoma and osteosarcoma, Cancer 62, 1691-1694 (1988). 31. Fearon, E. R., and Vogelstein, B. A genetic model for colorectal tumorigenesis, Cell 61, 759-767 (1990). 32. Fearon, E. R., Hamilton, S. R., and Vogelstein, B. Clonal analysis of human colorectal tumors, Science 238, 193-197 (1987). 33. Stenback, F. Benign, borderline and malignant serous cystadenomas of the ovary, Pathol. Res. Pratt. 172,58-72(1981). 34. Puls, L. E., Powell, D. E., DePriest, P. D., Gallion, H. H., Hunter, J. E., Kryscio, R. J., and van Nagell, J. R., Jr. Transition from benign to malignant epithelium in mutinous and serous ovarian cystadenocarcinoma, Gynecol. Oncol., in press. 35. Atkin, N. B., and Baker, M. C. Deficiency of all or part of chromosome 11 in several types of cancer: Significance of a reduction in the number of normal chromosomes 11, Cytogenet. Cell Genet. 47, 106-107 (1988). 36. Li, S., Schwartz, P. E., Lee, W. H., Yang-Feng, T. L. Allele loss at the retinoblastoma locus in human ovarian cancer, J. Natl. Cancer Inst. 83(9), 637-640 (1991). 37. Lee, J. H., Kavanagh, J. J., Wharton, J. T., Wildrick, D. M., and Blick, M. Allele loss at the c-Ha-rasl locus in human ovarian cancer, Cancer Res. 49, 1220-1222 (1989).
