Loss of heterozygosity on chromosome 17qll-21 in cancers of women who have both breast and ovarian cancer

Joellen M. Schildkraut, PhDT N. Keith Collins, MS,’ Georgette A. Dent, MD, J. Allen Tucker, MD,” J. Carl Barrett, PhD,” Andrew Berchuck, MD,b and Jeff Boyd, PhD

Durham, Research Triangle Park, and Chapel Hill, North Carolina, and Mobile, Alabama

OBJECTIVE: Our purpose was to determine the frequency of allele loss in the region of the BRCAI gene in cancers of women who have both breast and ovarian cancer. STUDY DESIGN: Four polymorphic microsatellite markers on chromosome 17ql l-21 were examined by the polymerase chain reaction in deoxyribonucleic acid from paraffin blocks of normal tissues, breast cancers, and ovarian cancers in 24 women who had primary cancers in both sites. RESULTS: Loss of heterozygosity was seen in one or more markers on chromosome 17qll -q21 in 46% of breast cancers and 78% of ovarian cancers. In 38% of cases allele loss was seen in both cancers, and in all these cases the same allele was lost in both cancers. Significantly younger ages at diagnosis of both breast and ovarian cancer were noted among cases with allele loss in both cancers compared with cases in which allele loss was found only in the ovarian cancer (p < 0.05). CONCLUSIONS: Because cases in which 17ql l-21 allele loss was seen in both cancers had a young age of onset and the same allele was always deleted in both cancers, hereditary alterations in BRCAl may play a role in this subset. The older age of onset in cases in which allele loss was seen only in the ovarian cancer suggests that the development of these cancers is not related to an inherited defect in BRCAl. (AM J OBSTET GYNECOL 1995;172:908-13.)

Key words: Ovarian cancer, breast cancer, chromosome 17q, loss of heterozygosity, BRCAl

Most human cancers are thought to arise from a multistep process in which growth regulatory genes are damaged. A striking similarity has been noted between the molecular alterations present in breast and ovarian cancers, which include amplification of the HER-P/neu, and c-myc oncogenes and mutation of the p53 tumor suppressor gene.’ In addition, several studies of breast and ovarian cancers suggest that a tumor-suppressor gene of importance may reside on chromosome 17q.2-’ First, linkage to a locus on 17q (BRCAl) has been demonstrated for early-onset breast cancer and in fami- lies with multiple cases of breast and ovarian cancer.“, 4 In addition, it has been shown that loss of heterozygos- ity on 17q frequently occurs in both familial and spo- radic breasP” and ovarian cancers.“-‘S In each of nine breast-ovarian cancer families loss of heterozygosity on

From the Cancer Prevention, Detectron and Control Research Pro-

g7um” and the Division of Gynecologic Oncology/Department of Ohstetws and Gynecology,” Duke Comprehensive Cancer Center, Duke University Medical Center, the Laboratory of Molecular Car- cinogenesis, National Institute of Environmental Health Sciences, National Institutes of Health,’ the Department of Pathology, Univer- sity of North Carolina School of Medicine,d and the Department of PatholoLq, University of South Alabama Medical Center.’ Received for bublication May 27, 1994; revised August 12, 1994; accepted ie&mber 9, 1994.. Retrint reauests: Andrew Berchuck, MD, Duke University Medical Ceiter, Boi 3079, Durham, NC 2?‘710. (i/1/6051 7

17q affected the wild-type chromosome, consistent with the hypothesis that the predisposing gene (BRCAl) is a tumor suppressor.‘* If this is the case, affected individu- als in familial breast-ovarian cancer kindreds presum- ably inherit one mutant copy of BRCAl, with tumor development dependent on subsequent loss of the sec- ond copy of the gene in a single cell. Loss of BRCAl might also be involved in the development of some sporadic breast and ovarian cancers. In this instance damage to both copies of BRCAl would be acquired rather than inherited. The observation that familial breast-ovarian cancer cases have a younger mean age of onse?. ” is consistent with this model in which only one additional event is required in familial cases, whereas two events are required in sporadic cases.16

Identification of the BRCAl gene could potentially have important implications for both our understand- ing of the pathogenesis of breast and ovarian cancer and for risk assessment. Currently several groups are working to identify the BRCAl gene by means of positional cloning. In the current study we determined the frequency of loss of heterozygosity on chromosome 17qll-2 1 in a group of breast and ovarian cancers from women who have independent primary cancers at both sites. The goal of the study was to ascertain what fraction of these cases had loss of heterozygosity on chromosome 17ql l-2 1 in one or both cancers sugges-

908

Volume 172, Number 3 Am J Obstet Gynecol

Schildkraut et al. 909

Al 2 3 4

B 0 N B 0 N BON BON

THRA 1

B 1 2 3 4

B 0 N B 0 N B 0 N B 0 N

B 0 N B 0 N B 0 N B 0 N 6 0 N

0178293 D17S588

Fig. 1. Demonstration of loss of heterozygosity on chromosome 17q by means of microsatellite polymorphisms (B, breast cancer DNA; 0, ovarian cancer DNA, N, normal DNA). A, THRAl -case 1, noninformative; case 2, no allele loss in either cancer; case 3, loss of lower allele in ovarian cancer; case 4, loss of upper allele in both cancers. B, D17S579 -case 1, noninformative; case 2, no allele loss in either cancer; case 3, loss of upper allele in ovarian cancer; case 4, loss of lower allele in both cancers. C, D175293-case 1, noninformative; case 2, loss of upper allele in both cancers. D, D17S588-case 1, no allele loss in either cancer; case 2, loss of lower allele in both cancers; case 3, loss of lower allele in ovarian cancer.

tive of a role for BRCAI. Conversely, we sought to define a subset of cases without allele loss on 17qll-21 in which BRCAl might not play a role.

Material and methods Tissue specimens. Tumor registries at Duke Univer-

sity Medical Center and the University of North Caro- lina at Chapel Hill were reviewed (1970 through 1989) to identify women who had independent primary can- cers in both the breast and ovary. During the years of this study 7166 cases of breast cancer and 1758 cases of ovarian cancer were treated at the participating hospi- tals. Fifty dual-primary cases were identified, and in 31 cases the breast cancer was the first primary, in 15 cases

the ovarian cancer was the first primary, and in four cases the breast and ovarian cancers were diagnosed concomitantly. We were able to obtain paraffin blocks from both primary cancers and adjacent normal tissue in 30 cases. Slides fi-om each block were reviewed by the study pathologist (G.A.D.) to verify the diagnoses and to classify the histologic type and grade. There was agreement with the original pathologic diagnosis for 29 of the 30 cases. In addition, the portion of the block with the greatest percentage of tumor was identified and the degree of stromal involvement was determined.

Molecular techniques. Isolation of genomic deoxy- ribonucleic acid (DNA) from formalin-fixed, paraffin- embedded tissue samples was performed as previously

910 Schildkraut et al. March 1993 Am J Obstet Gynecol

174 case# 470 200 460 380 390 410 10 300 290 190 440 260369 40 240 280 60 20 25 400 310 130 90 70 CM AGE BRCA 19 27 38 39 40 40 42 44 44 51 48 47 48 52 57 60 35 58 59 65 67 69 71 83

AGE O’JCA 20 39 39 46 39 46 49 46 50 44 49 51 49 50 51 51 53 58 62 71 67 74 71 73 34% I

Fig. 2. Loss of heterozygosity on chromosome 17q in breast and ovarian cancers. Left half of each circle contains results for breast cancer and right half contains results for ovarian cancer (solid, allele loss; hatched, no allele loss; open, noninformative; missing, not done).

described.” Oligonucleotide primers flanking four polymorphic (CA),, microsatellite loci on chromosome 17qll-21 were used in this study (Figs. 1 and 2).“, ” Polymerase chain reaction reactions consisted of the fol- lowing: 50 mmol/L potassium chloride, 10 mmol/L Tris (pH 8.3), 1.5 mmol/L magnesium chloride, 200 FmoliL each deoxyribonucleoside triphosphate, 1 IrmoliL each primer, 0.5 unit AmpliTaq polymerase (Perkin- Elmer/Cetus, Emeryville, Calif.), and 10 to 30 ng of genomic DNA in a volume of 20 ~.rl. One primer was end-labeled with y-phosphorus 32-labeled adenosine triphosphate by polynucleotide kinase, with the KinAce-It kit (Stratagene, La Jolla, Calif.) and column- purified before the polymerase chain reaction. Thirty cycles were performed, consisting of 30 seconds at 94” C, 30 seconds at 55” C, and 30 seconds at 72” C, followed by a ‘i-minute extension at 72” C with a Perkin-Elmer 9600 GenAmp polymerase chain reaction system. The polymerase chain reaction products were processed by diluting 1: 1 with a loading buffer consist- ing of 98% formamide, 10 mmol/L ethylenediaminetet- raacetic acid (pH 8.0), 0.02% xylene cyan01 FF, and 0.02% bromophenol blue and then were denatured for 2 minutes at 90” C. Typically, 5 ~1 of this solution was subjected to electrophoresis in 6% polyacrylamide gels containing 8.3 mol/L urea for 2 to 3 hours at 70 W. The gels were fixed in 10% methanol and 10% acetic acid, dried, and exposed to x-ray film at -80” C. For informative (heterozygous) cases allele loss was scored as a change of greater than twofold in the ratio of one allele to the other in cancer compared with normal tissue, as determined by densitometric analysis of the resultant autoradiogram.

Statistical analysis. Statistical analyses were per- formed with SAS software (SAS Institute, Gary, N.C.). A tumor was classified as having loss of heterozygosity on chromosome 17qll-21 if the DNA from the normal tissue was heterozygous for at least one of the markers from a given region of chromosome 17q and loss of one allele was observed in the tumor DNA. The Wilcoxon

rank-sum test was applied to compare differences in age at diagnosis. Fisher exact tests were used to com- pare differences in proportions.

Results Successful amplification of target DNA sequences on

chromosome 17ql l-21 was achieved in normal tissues in 24 of 29 cases in which paraffin blocks were available. Among these 24 cases results were also obtained for 22 of the breast cancers and 23 of the ovarian cancers, which included 2 1 of the first primary cancers and 24 of the second primary cancers. Twenty patients were white and four were nonwhite. The median age at diagnosis was 48 years for the breast cancers and 50 years for the ovarian cancers. The median age at which the first cancer was diagnosed in these 24 women was 47 years. In 15 cases the breast cancer was diagnosed first (me- dian age 44 years), in seven cases the ovarian cancer was diagnosed first (median age 51 years), and in two cases the two cancers were diagnosed concomitantly. The average time between diagnosis of the first and second primary tumors was 4.6 + 3.9 years. The relationship between 17ql l-2 1 allele loss and clinicopathologic fea- tures is shown in Table I. The only significant finding was that epithelial ovarian tumors of low malignant potential had a lower frequency of allele loss (40%) compared with invasive cancers (89%) (fi = 0.04).

At least one of the four polymorphic markers was heterozygous in DNA from normal tissue in each of the 24 cases. Representative examples of chromosome 17qll-21 allelic loss are demonstrated in Fig. 1, and the results for all four markers are shown in Fig. 2. Loss of one allele of one or more of the markers was seen in 10 of 22 (46%) breast cancers and 18 of 23 (78%) ovarian cancers. There was no significant difference between the frequency of loss of various markers in the breast or ovarian cancers. There was no difference in the frequency of 17ql l-2 1 allele loss between the breast cancers that occurred first, seven of 13 (54%), and breast cancers that occurred after the diagnosis of

Volume 172, Number 3 ilm J Obstet Gym01

Schildkraut et al. 911

Table I. Relationship between clinicopathologic characteristics and loss of heterozygosity on chromosome

l’cl

Breast cancers* Ovarian cancers

No/total % No./total %

Histologic features In situ Infiltrating ductal Medullary

Grade Low malignant potential Well differentiated Moderately differentiated Poorly differentiated

Stage Confined to breast Metastatic

Ill 8120 l/l

- 112 o/5 s/14

419 5112

100 40

100

50 0

57

44 42

serous 16/19 84 mutinous O/2 0 endomet. l/l 100 undiff. l/l 100

215 40t 414 100 718 88 516 83

I/II 518 63 III/Iv 13115 87

*Information on grade and stage not available in one breast cancer case. tp = 0.04 (low malignant potential vs other grades).

Table II. Relationship between age at cancer diagnosis and loss of heterozygosity on chromosome 17q

Breast cancers -Age at diagnosis (yr) Ovarian cancers-Age at diagnosis

17q allele loss” No. of pain Mean SD Mean SD

Neither cancer 3 47.3 32.6 44.0 26.9 Breast cancer 2 47.5 13.4 44.5 9.2 Ovarian cancer 8 59.4 10.4 61.1 11.0 Both cancers 8 45.3 9.7 48.6 6.7

*Age at breast and ovarian cancer diagnosis is younger in cases with 17q allele loss in both cancers compared with cases with allele loss in ovarian cancer only (p < 0.05).

ovarian cancer, three of seven (43%). Allele loss was both cancers (p < 0.05). Although women who had seen in three of six (50%) of ovarian cancers that were allele loss in the breast cancer only or in neither cancer the first primary and in a somewhat higher fraction of also appeared to have an early age of onset of the first cases in which the ovarian cancer was the second pri- cancer, the small number of cases in these subsets did mary, 13 of 15 (87%). not allow for meaningfU1 statistical analysis.

Among the 24 women in the study both cancers were successfully analyzed in 2 1 cases. Allele loss on chromo- some 17qll-21 was seen in both cancers in eight of these 21 cases (38%), including six of 13 (46%) cases in which the breast cancer occurred first, two of six (33%) cases in which the ovarian cancer occurred first, and none of two cases that occurred concomitantly. In all eight cases in which 17ql l-2 1 allele loss occurred in both the breast and ovarian cancers the same allele was lost in both cancers (Fig. 1). Finally, allele loss was seen only in the breast cancer in two cases, only in the ovarian cancer in eight cases, and in neither cancer in three cases.

Comment

The relationship between age at diagnosis and allele loss on chromosome 17qll-21 among various subsets of patients is demonstrated in Table II. In cases in which allele loss was seen only in the ovarian cancer, cancers were diagnosed at a significantly older age compared with cases in which allele loss was seen in

The aim of this study was to determine whether the BRCAI putative tumor suppressor gene might be in- volved in the development of breast and ovarian can- cers in women who have primary cancers in both sites. The young median age at diagnosis of ovarian and breast cancers in this series (approximately 45 to 50 years) relative to sporadic cases (approximately 55 to 60 years) is supportive of the hypothesis that these patients have a hereditary predisposition to cancer. A similar young median age of cancer diagnosis was noted in the families with multiple cases of breast and ovarian cancer that were used for the original studies that found linkage to chromosome 17q (BRCA1).3

Because the young age of onset in women with cancers in both the breast and ovary suggests the possibility of a hereditary predisposition, we sought to determine whether the BRCAl gene might play a role

912 Schildkraut et al. March 1995 AITI J Obstet Gynecol

in the pathogen&s of these cancers. If BRCAl acts in a manner similar to other tumor suppressor genes, in addition to inheritance of one mutant copy of the gene development of cancer at a young age is likely to be dependent on inactivation of the second copy of the gene. In view of this, until the BRCAl gene is identified, allele loss on chromosome 17qll-21 serves as an indi- rect marker for potential involvement of BRCAI. It is possible, however, that allele loss on 17q may occur in spite of the presence of a normal copy of BRCAl on the other chromosome. In this regard, our group and others have found that most of the deletions on 17q in ovarian cancers involve loss of the entire chromosomal arm.” Thus allele loss on 17q may reflect generalized genomic instability rather than specific inactivation of the BRCAl tumor suppressor gene. Conversely, in cases in which allele loss is not seen on 17ql l-2 1 it is possible that small deletions in the area of BRCAl could be missed because the markers used do not overlap this gene.

By means of highly informative microsatellite markers we determined that 46% of the breast cancers and 78% of the ovarian cancers in this study had loss of heterozygos- ity on chromosome 17qll-21. Similarly, our group and others found a high frequency of allele loss on 17ql l-2 1 in sporadic breast and ovarian cancers.“.“. ” In addition, as shown previously, ovarian tumors of low malignant potential had a lower frequency of allele loss than did invasive ovarian cancers.” Allele loss was seen in both cancers in 38% of cases, and in all of these cases the same allele was lost in both the breast and ovarian cancer, consistent with the hypothesis that deletion of the re- maining normal allele occurs, leaving only the inherited mutant allele. Allele loss was seen in the ovarian cancer only in 38% of cases, and in a few cases allele loss was seen in only the breast cancer or in neither cancer.

The finding that the pattern of chromosome 17ql l-2 1 allele loss in the breast and ovarian cancers defines subsets of cases on the basis of age at diagnosis is supportive of the hypothesis that inherited mutations in BRCAl play a role in the development of some of these cancers. Because cases in which allele loss was seen in both cancers had a young age at diagnosis and the same allele was always deleted in both cancers, a role for BRCAl would seem most likely in this subset. Although a young age at diagnosis was also seen in cases with allele loss in only the breast cancer or in neither cancer, the number of cases in these subsets was too small to allow meaningful analysis. It is possible that BRCAl also is involved in the development of these cancers but that small deletions were not detected with the markers used in this study. In addition, it is possible that, like the p53 tumor suppressor gene, BRCAl may act as a dominant negative oncogene, in which case loss of the second allele might not be requisite for tumori-

genesis. Alternatively, the young age at diagnosis in these subsets may reflect the inheritance of a defective gene distinct from BRCAl.

The older ages at diagnosis of breast and ovarian cancers in the cases in which allele loss was seen only in the ovarian cancer suggests that the development of these cancers is not related to a hereditary predisposi- tion. In this regard, a small number of women would be expected to coincidentally have sporadic cancers in both the breast and ovary. Because of the higher inci- dence and cure rate (70%) of breast cancer relative to ovarian cancer (30%), there are more women in the population with breast cancer who are at risk for ovar- ian cancer than there are women with ovarian cancer who are at risk for breast cancer. Consistent with this, overall we found that breast cancer was diganosed first in 15 cases, whereas ovarian cancer was diagnosed first in only seven cases. Similarly, in the subset of cases in which allele loss was seen only in the ovarian cancer the first diagnosis was breast cancer in five cases and ovar- ian cancer in one case, and the two cancers were diagnosed simultaneously in two cases.

When the BRCAl gene is identified, it is likely that techniques for detection of altered forms of the gene will play a role in screening families at high risk for early-onset breast and ovarian cancer. Individuals with a defective gene could then be subjected to intensive surveillance with screening modalities such as radio- graphic studies and serum tumor markers. In addition, prophylactic subcutaneous mastectomy and prophylac- tic oophorectomy might be strongly recommended for women demonstrated to carry the altered form of BRCAl. Currently, because the BRCAl gene remains elusive, it is difficult to counsel individual patients, because it is not possible to distinguish affected indi- viduals from unaffected. This study suggests that women who have independent primary cancers in the breast and ovary can be divided into subsets, however. Cases in which 17qll-2 1 allele loss occurs in the ovar- ian cancer only, generally, have a late age of onset, and there is no evidence to support a hereditary predispo- sition. In contrast, early age of onset and chromosome 17ql l-2 1 allele loss in both cancers suggests a role for inherited alterations in BRCAl. Perhaps, until the BRCAl gene is discovered, the finding of 17qll-21 allele loss in both cancers might warrant recommending that other women in these families undergo intensive surveillance or prophylactic surgery.

Addendum After this article was accepted for publication, the

BRCAl gene on chromosome 17q was identified. We plan to examine the DNA samples used in this study from women who have both breast and ovarian cancer for mutations in the BRCAl gene.

Volume 172, Number 3 Am J Obstet Gynecol

Schildkraut et al. 913

1.

2.

3.

4.

5.

6.

REFERENCES 10.

Berchuck A, Bast RC. Growth factors, oncogenes, and tumor-suppressor genes. In: Markman M, Hoskins WJ, eds. Cancer of the ovary. New York: Raven Press, 1993: 61-78. 11.

Hall JM, Lee MK, Newman B, et al. Linkage of early-onset familial breast cancer to chromosome 17q21. Science 1990;250:1684-9. Narod SA, Feunteun J, Lynch HT, et al. Familial breast- ovarian cancer locus on chromosome 17q12-q23. Lancet 1991;338:82-3.

12.

Easton DF, Bishop DT, Ford D, Crockford GP, Breast Cancer Linkage Consortium. Genetic linkage in familial breast and ovarian cancer: results from 214 families. Am J Hum Genet 1993;52:678-701. Hall JM, Friedman L, Guenther C, et al. Closing in on a breast cancer gene on chromosome 17q. Am J Hum Genet 1992;50:1235-42. Sato T, Futoshi A, Sakamoto G, Nakamura KY. Accumu- lation of genetic alterations and progression of primary breast cancer. Cancer Res 1991;51:5794-9. Sato T, Saito H, Swenson J, et al. The human prohibitin gene located on chromosome 17q21 is mutated in spo- radic breast cancer. Cancer Res 1992;52:1643-6. Futreal PA, Soderkvist P, Marks JR, et al. Detection of frequent allelic loss on proximal chromosome 17q in sporadic breast carcinoma using microsatellite length polymorphisms. Cancer Res 1992;52:2674-7. Cropp CS, Lidereau R, Callahan R. Loss of heterozygosity of chromosomes 17 and 18 in primarv breast cancers. Proc Nat1 Acad Sci U S A 1990:87:7737-41.

13.

14.

15.

16.

17.

18.

Sato 7; Tanigami A, Yamakwa K, Kasumi AF, Sakamoto G, Nakamura Y. Allelotype of breast cancer: cumulative allele losses promote tumor progression in primary breast can- cer. Cancer Res 1990;50:7184-9. Jacobs IJ, Smith SA, Wiseman RW, et al. A deletion unit on chromosome 17q in epithelial ovarian tumors distal to the familial breast/ovarian cancer locus. Cancer Res 1993;53: 1218-21. Foulkes W, Blasck D, Solomon E, Trowsdale J. Allele loss on chromosome 1’7q in sporadic ovarian cancer. Lancet 1991;338:444-5.

Eccles DM, Cranston G, Steel CM, Nakamura Y, Leonard RCF. Allele losses on chromosome 17 in human epithelial ovarian carcinoma. Oncogene 1990;5:1599-601. Smith SA, Easton DF, Evans DGR, Ponder BAJ. Allele losses in the region 17q12-21 in familial breast and ovar- ian cancer involve the wild-type chromosome. Nature Genet 1992;2:128-31. Bewtra C, Watson P, Conway T, Read-Hippee C, Lynch HT. Hereditary ovarian cancer: a clinicopathological study. Int J Gynecol Path01 1992;11:180-7. Knudson AG Jr. Hereditary cancer, oncogenes, and anti- oncogenes. Cancer Res 1985;45:1437-43. Sasaki H, Nishii H, Takahashi H, et al. Mutation of the Ki-ras protooncogene in human endometrial hyperplasia and carcinoma. Cancer Res 1993;53:1906-10. Anderson LA, Friedman L, Osborne-Lawrence S, et al. High-density genetic map of the BRCAl region of chro- mosome 17q12-21. Genomics 1993;17:618-23.

Make researching easier with the Ten-Year Cumulative Index (1978-1987) to the AMERICANJOURNALOF OBSTETRICSANDGYNECOLOGY

The Ten-Year Cumulative Index is a complete guide to more than 30,000 pages of original articles, case reports, letters, and editorials published in the AMERICAN JOURNAL OF OBSTETRICS AND GYNECOLOGY from 1978 through 1987 (volumes 130 to 157). Num- bering more than 800 pages, the Cumulative Index includes both a subject and an author index. This hard-cover volume is $64.50 ($69.00 international). Prices include shipping. Payment must accompany all orders and must be in U.S. funds, &awn on a U.S. bank.

Contact Mosby-Year Book, Inc., Subscription Services, 11830 Westline Industrial Drive, St. Louis, MO 63146-3318, USA; phone (314)453-4351. In the U.S., call toll-free: (800)453-4351. Please allow six weeks for delivery.