BRCA1/2 Mutation Status Influences Somatic Genetic Progression in Inherited and Sporadic Epithelial Ovarian Cancer Cases1

[CANCER RESEARCH 63, 417–423, January 15, 2003] BRCA1/2 Mutation Status Influences Somatic Genetic Progression in Inherited and Sporadic Epithelial Ovarian Cancer Cases1

Susan J. Ramus,2 Paul D. P. Pharoah, Patricia Harrington, Carole Pye, Bruce Werness, Lynda Bobrow, Ayse Ayhan, Dagan Wells, Ami Fishman, Martin Gore, Richard A. DiCioccio, M. Steven Piver, Alice S. Whittemore, Bruce A. J. Ponder, and Simon A. Gayther3 Department of Oncology, Strangeways Research Laboratories, Cambridge, United Kingdom [S. J. R., P. D. P. P., P. H., C. P., B. A. J. P.]; Department of Pathology, Inova Fair Oaks Hospital, Fairfax, Virginia [B. W.]; Department of Pathology, Addenbrookes Hospital, Cambridge, United Kingdom [L. B.]; Cancer Research United Kingdom Translational Oncology Laboratory, John Vane Science Centre, Charterhouse Square, London, United Kingdom [A. A., S. A. G.]; University College London, Department of Obstetrics and Gynaecology, London, United Kingdom [D. W.]; Department of Obstetrics and Gynaecology, Meir Hospital, Sapier Medical Centre, Meyir Hospital, Kfar-Saba, Israel [A. F.]; Department of Clinical Oncology, Royal Marsden Hospital, London, United Kingdom [M. G.]; Department of Cancer Genetics Roswell Park Cancer Institute Buffalo, New York [R. D., M. S. P.]; and Department of Health Research and Policy, Stanford University School of Medicine, Stanford, California [A. S. W.]

ABSTRACT of serous cystadenocarcinomas in familial ovarian cancers (89%) compared with nonfamilial tumors (49%; Ref. 10). Other reports also Metaphase comparative genomic hybridization was used to analyze the indicate that BRCA1 tumors are mostly serous adenocarcinomas (9, spectrum of genetic alterations in 141 epithelial ovarian cancers from 11). Finally, some studies suggest that BRCA1/2 mutation status may BRCA1 and BRCA2 mutation carriers, individuals with familial non- BRCA1/2 epithelial ovarian cancer, and women with nonfamilial epithe- influence patient survival, although there is disagreement between lial ovarian cancer. Multiple genetic alterations were identified in almost studies as to whether BRCA1 carriers have a better or worse survival all tumors. The high frequency with which some alterations were identi- than controls (10–13). fied suggests the location of genes that are commonly altered during Taken together, these data suggest that BRCA1/2 mutation status ovarian tumor development. In multiple chromosome regions, there were may influence the clinical characteristics and outcome in breast and significant differences in alteration frequency between the four tumor ovarian cancers. However, mutations in these genes alone are unlikely types suggesting that BRCA1/2 mutation status and a family history of to account for all of the variation observed; tumor formation results ovarian cancer influences the somatic genetic pathway of ovarian cancer from an accumulation of somatic genetic alterations in several differ- progression. These findings were supported by hierarchical cluster anal- ent genes, of which many may influence tumor phenotype. In support ysis, which identified genetic events that tend to occur together during of this, some studies have shown association between the clinical tumorigenesis and several alterations that were specific to tumors of a particular type. In addition, some genetic alterations were strongly asso- characteristics of tumors and multiple differences in gene expression ciated with differences in tumor differentiation and disease stage. Taken (14). together, these data provide molecular genetic evidence to support previ- Previously, we have established the BRCA1 and BRCA2 mutation ous findings from histopathological studies, which suggest that clinical status in 288 epithelial ovarian cancer families (Ref. 3; unpublished features of ovarian and breast tumors differ with respect to BRCA1/2 data). The purpose of this study was to establish whether the spectrum mutation status and/or cancer family history. of somatic genetic events, which may influence tumor phenotype during ovarian cancer development, differs with respect to BRCA1/2 mutation status and/or a family history of the disease. To do this, we INTRODUCTION have compared the frequencies of genomic alterations identified using 4 Mutations in the BRCA1 and BRCA2 genes are responsible for metaphase CGH between ovarian tumors from BRCA1 and BRCA2 about half of all families containing two or more cases of epithelial mutation carriers, familial cases in which no BRCA1/2 mutation could ovarian cancer in close relatives and most families in which multiple be identified and sporadic cases. cases of ovarian and breast cancer occur together (1–4). Ovarian and breast tumors from mutation carriers frequently show loss of the MATERIALS AND METHODS normal allele (detected by microsatellite analysis), suggesting that both genes behave as tumor suppressors (5, 6). Patient Material. Paraffin-embedded epithelial ovarian tumors from 141 Several studies indicate that the histological and clinical features of individuals were analyzed in this study. Of these, 108 cases were from families ovarian and breast cancers vary with respect to BRCA1/2 mutation containing two or more first-degree relatives with ovarian cancer, identified status and a family history of the disease. In breast cancer, histopatho- from the United Kingdom Coordinating Committee on Cancer Research (3) logical characteristics such as grade, proliferation rate, S-phase frac- and United States Gilda Radner (15) familial ovarian cancer registries. All tion, and mitotic and aneuploid indices may differ either between cases have been analyzed for germ-line mutations throughout the coding region and splice site boundaries of the BRCA1 and BRCA2 genes (Ref. 3; BRCA1 and BRCA2 tumors or when compared with non-BRCA1/2 unpublished data). Mutations in BRCA1 were present in 46 individuals from tumors (7–9). In ovarian cancer, one study found a higher proportion 32 families and in BRCA2 in 18 individuals from 11 families. Also included in the study were five tumors from a hospital-based collection of ovarian Received 6/6/02; accepted 11/13/02. cancer cases from Israel, which carried the Ashkenazi Jewish founder mutation The costs of publication of this article were defrayed in part by the payment of page (6174delT) in BRCA2 (13). No identifiable BRCA1/2 mutation was present in charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 44 cases from 32 families. Sporadic tumors (i.e., cases with no reported first or 1 This research was supported by a program grant from Cancer Research United second-degree relatives with ovarian cancer) from 28 individuals, who had Kingdom, National Cancer Institute Research Grants R01CA66190 and 16056, and a taken part in a hospital-based collection of ovarian cancer cases at the Royal BRIEF Award from Brunel University. B. A. J. P. is a Glibb Fellow of Cancer Research Marsden Hospital (London, United Kingdom) were also analyzed. These cases United Kingdom 2 Present address: Department of Pathology, The University of Melbourne, Victoria, have previously been analyzed for germ-line mutations in the BRAC1 gene; Australia. none were identified (16). 3 To whom requests for reprints should be addressed, at Cancer Research United For each tumor, histological sub-type and tumor grade was established by Kingdom Translational Oncology Laboratory, Barts and the Royal London, Queen Mary School of Medicine and Dentistry, John Vane Science Centre, Charterhouse Square London EC1M 6BQ, United Kingdom. Phone: 44-20-7882-5795; Fax: 44-20-7882-6110; 4 The abbreviations used are: CGH, comparative genomic hybridization; LOH, loss of E-mail: [email protected]. heterozygosity. 417

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2003 American Association for Cancer Research. CGH ANALYSIS OF FAMILIAL VERSUS SPORADIC OVARIAN TUMOR specialist pathology review (B. W., L. B., A. A.). Information on disease stage, regions in common between tumors. In some cases it was possible to where available, was obtained from hospital records. Tumors from sporadic define more than one region of interest (Table 1). Seemingly, some cases were selected so that the distribution of different histological subtypes alterations extend for most of the length of a chromosome suggesting was similar to those in familial cases. In total, 59.5% of tumors were of serous a change in chromosome number rather than an interstitial alteration; histology (ranging from 52–63% between the four groups), 10.5% of tumors these include loss of chromosomes 4, 13, 14, 15, and X and gain of were endometrioid histology (range, 5–14%), 18% were either undifferentiated chromosomes 1, 12, 19, and 20. We cannot say whether changes in tumors or carcinomas of unspecified histology (range, 14–26%), and the remainder (12%; range, 9–18%) were a mixture of rare histological subtypes, chromosome number and interstitial alterations on the same chromo- including tumors of mucinous, clear cell, and mixed histology. some represent the same gene targets. Metaphase CGH Analysis. Pathology examination of tissue sections iden- High levels of amplifications were identified at several sites tified areas of tumor epithelium, which were microdissected away from sur- throughout the genome. In some instances, the same region of rounding tissues. DNA was extracted after proteinase K digestion. To improve high-level amplification was common to multiple tumors, which the sensitivity of metaphase CGH analysis, only regions of tissue containing possibly indicates the location of a single gene target. In general, Ն70% tumor epithelium were dissected. The same normal DNA sample, high-level amplifications were not frequent events. However, they which was extracted from normal archival tissue, was used as the reference for tended to occur in regions that also showed frequent gain, which all of the CGH analyses. may suggest a shared target (Table 2). The most frequent region of Whole genome amplification was performed using degenerate oligonucleo- high-level amplification, which was identified in 40% of tumors, tide priming PCR; DNA fragments were labeled either with biotin 16-dUTP (normal DNA), or digoxigenin-11-dUTP (tumor DNA; Roche). After ampli- occurred on chromosome 8q22-qter. Putative candidate genes in fication, tumor and normal samples were combined with 20 ␮g of human regions of amplification and in regions of frequent loss and gain CoT-1 DNA, denatured, and allowed to preanneal before hybridizing to (occurring in Ͼ50% of tumors) are listed in Table 2. These genes normal, male metaphase chromosome spreads for 72 h at 37°C. After hybrid- were selected because they have all previously been shown to have ization, biotin-labeled DNA was detected using avidin-Texas Red (Vector a role in cancer. Laboratories) and digoxigenin-labeled DNA detected using antidigoxigenin- Different Patterns of Somatic Alteration Associated with fluorescein Fab fragments (Roche). Slides were counterstained with 4Ј,6- BRCA1/2 Mutation Status and a Family History of Ovarian diamidino-2-phenylindole to identify the chromosomes. Slides were viewed on Cancer. We used two different approaches to determine whether a Zeiss fluorescence microscope. Digital images of metaphase spreads were the pattern of somatic genetic alterations during tumor develop- captured using Quips SmartCapture VP and analyzed using Quips CGH ment differs between BRCA1, BRCA2, familial non-BRCA1/2, Kayotyper and Interpreter software (Vysis). For each sample, at least 10 different metaphase chromosome spreads were and sporadic ovarian cancers. Firstly, we performed a systematic examined and a mean value for the ratio of the green signal (tumor DNA) to comparison of the frequency of genetic alterations between the red signal (normal DNA) calculated along the length of each chromosome. To four tumor groups. Secondly, we performed hierarchical cluster establish threshold levels for the recording of loss and gain in tumors, we analysis to identify alterations that tended to occur together during performed metaphase CGH analyses using two different normal DNA samples; tumor development. For the purpose of these analyses, we divided this provided an indication of the normal variation in the ratios of the two the genome into 100 nonoverlapping regions of similar size based fluorochromes for each chromosome. No genetic variation identified in nor- on the 4Ј,6-diamidino-2-phenylindole-banding on the CGH profile mal-normal CGH comparisons reached the thresholds for loss or gain that were and recorded the presence of loss or gain at every region for each subsequently used for the analysis of tumor samples. The threshold for record- tumor. ing genomic gain was a green to red ratio Ͼ 1.2, for genomic loss a green to We compared the frequency of alteration at all regions by tumor red ratio Ͻ 0.85, and for amplification a green to red ratio Ͼ 1.5. ␹2 Metaphase CGH Control Analyses. The following controls were per- type using an overall test on 3 degrees of freedom. Where the ␹2 Ͻ formed to test the efficacy of metaphase CGH data: (a) CGH was repeated on statistical significance of the overall was P 0.2 and the frequency a random series of 10 samples to ensure that the spectrum of alterations of the alteration was Ն15% for at least one tumor group, we carried detected in each tumor sample was reproducible. (b) CGH was repeated on 15 out pair-wise comparisons of the frequency of the change; each tumor samples using reverse labeling (normal samples labeled with digoxigenin, group was compared against each other and against the other tumor tumor samples with biotin). In all cases, the data established for inverse and groups combined (see Table 3). standard CGH analyses were concordant. (c) Thirty-five samples were ana- Significant differences in the frequency of loss or gain between one lyzed by LOH microsatellite analysis to establish concordance/discordance or more groups of tumor were identified at 20 different chromosome with deletions identified by CGH. Microsatellite markers at nine regions on regions; in total, 41 significant differences were observed between the different chromosome arms were informative for 83/95 deletions identified by four groups (Table 3). Notably, deletions of the region containing the CGH; LOH and CGH data were concordant for 76/83 deletions. BRCA1 gene (17q12-21) were significantly more frequent in tumors from BRCA1 mutation carriers than in non-BRCA1 tumors ϭ RESULTS (P 0.014). Similarly, deletions of the region containing the BRCA2 gene (13q12-13) were significantly more frequent in tumors from Frequency of Somatic Alterations. Metaphase CGH identified BRCA2 mutation carriers than in non-BRCA2 tumors (P ϭ 0.006). multiple somatic alterations in 137 of the 141 tumors. On average, This is consistent with data from LOH studies, which indicates that 21.4 alterations (95% confidence interval, 19.9–22.9) were identified deletion of the wild-type BRCA1 or BRCA2 allele is a frequent and per tumor. No significant differences were observed in the number of nonrandom event in tumors from mutation carriers. genetic alterations identified between BRCA1, BRCA2, familial non- We performed a hierarchical clustering algorithm, implemented in BRCA1/2, and sporadic tumors. the program Cluster (17) to distinguish the most critical changes of Table 1 summarizes the frequencies with which genetic alterations tumor progression for each of the four tumor types. Cluster analysis were identified for each chromosome arm. Several alterations oc- was performed separately for each tumor type; this was necessary curred with a particularly high frequency (in 42–76% of all tumors); because many of the more frequent genetic alterations were common these were loss on chromosomes 4q, 5q, 6q 13q, 18q, and X and gain to all four groups and a cluster analysis of all tumors grouped together on chromosomes 1, 3q, 6p, 7q, 8q, 19q, and 20q. Several regions of could not differentiate between the groups. These data are illustrated relatively frequent loss or gain (in Ͼ30% of all tumors) were also in Fig. 1. For each group, two clusters representing regions of loss and identified. For most alterations, we were able to define single, critical gain were prominent. As expected, there was a tendency for some of 418

Table 1 Frequencies of loss and gain for each chromosome arma Genomic loss Genomic gain

Chromosome BRCA1 BRCA2 Non-1/2 Sporadic Total Critical BRCA1 BRCA2 Non-1/2 Sporadic Total Critical arm (%) (%) (%) (%) (%) regions (%) (%) (%) (%) (%) regions 1p 17 17 18 25 19 39 61 53 61 51 pter-p32 1q 15 13 5 7 10 43 61 56 50 52 cen-q22; q31-qter 2p 0 5 0 4 2 26 43 34 32 33 pter-p14 2q 28 22 23 32 26 q21-q32 29 26 16 21 23 q33-qter 3p 39 43 39 36 39 pter-p22; p13-cen 9 13 16 29 16 3q 11 9 9 18 11 52 56 43 43 48 q21-qter 4p 37 34 41 39 38 p14-cen 9 17 18 11 13 4q 74 91 70 75 76 cen-q31.3 7 5 3 4 5 5p 11 17 14 11 13 44 39 32 25 35 pter-p14 5q 78 74 73 54 71 q13-q22 17 31 12 14 17 q32-qter 6p 9 5 7 14 9 52 53 45 36 47 pter-p22 6q 80 83 70 68 75 q21-q23 2 9 7 4 5 7p 26 39 23 25 27 pter-p15 11 9 11 14 11 7q 7 9 7 11 8 52 48 52 43 50 q21-qter 8p 37 39 43 25 37 pter-p21 16 0 7 14 10 8q 7 13 5 4 6 83 78 64 64 72 q23-qter 9p 41 43 30 43 38 pter-p21 24 9 9 11 14 9q 26 43 7 50 28 q21-q22 26 31 45 25 33 q31-qter 10p 9 5 5 0 5 31 922 32 25 pter-p13 10q 15 22 23 18 19 17 35 18 25 22 q24-qter 11p 24 31 27 14 24 pter-p14 9 0 6 11 7 11q 22 9 23 25 21 45 53 32 18 37 cen-q13; q23-qter 12p 13 13 7 11 11 28 34 27 36 30 pter-p12 12q 43 22 34 21 33 q21-q23 31 34 23 36 30 q24.1-qter 13q 59 87 55 64 63 cen-q14; q21-q22 997 06 14q 30 48 27 39 34 cen-q13 24 26 12 32 22 q24-qter 15q 30 43 39 36 36 cen-q15; q21-q23 33 43 14 29 28 q24-qter 16p 4 0 11 4 6 21 22 14 14 18 16q 30 31 43 39 36 cen-q21 9 0 7 4 6 17p 39 34 32 11 30 17p 2 5 4 4 4 17q 41 17 27 18 28 cen-q21 31 34 26 29 29 q22-qter 18p 9 0 7 14 9 5 30 16 21 16 18q 59 61 52 46 55 q21-qter 9 17 12 11 12 19p 2 0 2 7 3 29 31 39 46 36 19p 19q 4 0 0 0 1 43 53 50 54 49 19q 20p 0 0 5 0 1 39 39 28 39 36 20p 20q 0 0 0 0 0 66 78 61 68 67 20q 21q 28 35 18 21 25 21q 31 39 23 29 29 21q 22q 24 22 25 14 22 17 17 19 43 23 22q Xp 37 43 36 57 42 Xp 11 5 3 4 6 Xq 43 58 57 54 52 q21-qter 22 17 14 14 17 a The frequencies of loss or gain were established from the number of chromosome arms that had either type of genetic alteration. Frequencies Ն 30% are highlighted in bold. Critical regions of loss and gain were determined only for alterations with a frequency of Ն 30% for one or more of the four tumor groups. the more frequent somatic alterations to cluster together, regardless of total) and on disease stage for 65 tumors. In general, there were fewer BRCA1/2 mutation status or family history. Regions of loss that were genetic alterations in grade 1 tumors (mean, 15.5) than in grade 2 or common to clusters from three or more groups occurred on chromo- grade 3 tumors (mean, 22.6 and 21.6 respectively), but this difference somes 4q, 5q, 6q, 13q, and 18q; common regions of gain occurred on was not statistically significant. Neither was there a significant dif- chromosomes 1p, 3q, 6p, 8q, 19, and 20. ference in the number of genetic alterations identified in stage I/II In addition to the frequently occurring alterations, there were sev- (21.1) compared with stage III/IV (21.9) tumors. eral changes that were not common in the clusters of all tumor types, We used ␹2 tests to compare the frequencies of loss and gain which may indicate differences in the molecular genetic pathways of throughout the genome with tumors of different grade and stage. In tumor progression between different tumor types. Significantly, loss addition, a ␹2 test for trend was used to assess whether there was a of the BRCA1 region appeared in the most prominent cluster in tendency for somatic alterations either to increase or decrease in BRCA1 tumors but not in familial non-BRCA1/2 or sporadic tumors. frequency from grade 1 to grade 3 tumors. These data are summarized Interestingly, loss at BRCA1 also appears in the most prominent in Table 4. Several alterations differed in frequency between tumors cluster in BRCA2 tumors, even though loss at this locus occurred in of different grade and/or stage. For example, loss at Xcen-q13 was only 18% of this tumor type. Other alterations identified in the clusters more common in grade 3 tumors compared with grade 1 and 2 tumors of one or two, but not all, tumor groups include gains on chromosome ϭ 5p and 7q (in BRCA1 and familial non-BRCA1/2 tumor clusters (P 0.002). Loss at 5q14-q22 was more common in grade 2 and 3 ϭ only), loss on X (BRCA2 and sporadic tumors), loss on 17p (BRCA1 tumors compared with grade 1 tumors (P 0.045), but an increasing and BRCA2 tumors), loss on 3p and gain on 11q (BRCA2 tumors), trend in frequency from grade 1 to grade 3 tumors was more signif- ϭ gain on 2p (familial non-BRCA1/2 tumors), and loss on 9q and gain icant (P 0.018). Loss on chromosome 5q14-q22 was also associated on 21q (sporadic tumors). with later stage disease (P ϭ 0.032) as were gains on chromosomes Association between Somatic Alterations and Clinical Charac- 21q22-qter (P ϭ 0.027) and 1p32-p34.3 (P ϭ 0.04). Surprisingly, two teristics of Ovarian Tumors. To assess the clinical significance of alterations showed a decrease in frequency with higher grade or later somatic changes in ovarian cancer, we classified tumors according to stage; a gain on 22q12-qter was more frequent in tumors of lower histopathological grade and disease stage. Information on grade was grade (P ϭ 0.029; ␹2 trend P ϭ 0.005), and a loss on 4p15.1-pter was available for all except three clear cell carcinomas (138 tumors in associated with earlier stage (P ϭ 0.005). 419

Table 2 Cancer-associated genes and putative targets of located within the most frequent regions of loss, gain and amplification identified in this studya Frequency Chromosome Region Alteration typeb (% of all tumorsc) Cancer associated genes within regiond 1p pter-p32 Gain 52 EPS15; M1S1; BLYM; JUN; RLF; TGFBR3; DDIT1; TIE1; MYCL1; MPL; LCK; EBVS1 PLA2G2A; RAP1; GA1; FGR; TNFR2; NB; HKR3; BRCD2; CDC2L1; E2F2; MLM; RIZ; MTS1; ERPL1 1q q21 Amplification 5 (33%) NTRK1; PE1; MUC1; PRCC 1q cen-q22; q31-qter Gain 51 NTRK1; PE1; MUC1; PRCC; SKI; PDGFA; PCTK3; ELK4; RBBP5; TGFB2; HPC2 2p p21 Amplification 2 (14%) FCC1 3q q26.1-qter Amplification 8 (44%) TERC; PIK3CA; ECT2; EVI1; MDS1; BCL6; EIF4G; FGF12 4q cen-q31.3 Loss 77 GRO1; GRO2; GRO3; IGFBP7; KDR; KIT; PDGFRA; AREG; RAP1; GDS1; FGF5; INP10; FGF2; EGF; IL2; CCNA; TYS; HCC2 5q q13-q22 Loss 71 CCNH; RASA1; APC; MCC 6q q21-q23 Loss 76 CCNC; FYN; MYB; ROS1; CTGF 8q q12 Amplification 7 (19%) MOS 8q q22-qter Amplification 40 (70%) MYC; CBFA2T1; MYBL1; NOV; PSCA; GLI4; GML; PVT1 Gain 72 10q q25-q26 Amplification 4 (20%) FGFR2; FGF8; DMBT1; MXI1; PRCA1; DEC; 11q q13 Amplification 2 (25%) FGF3; CCND1; FGF4; EMS1; BCL1; THRSP; FOSL1; MEN1; PPP1A; SEA; ST3 12p p11.2 Amplification 3 (11%) KRAG, RBBP2 13q cen-q14; q21-q22 Loss 60 BRCD1; FGF9; BRCA2; FLT1; RB1 18q q21-qter Loss 55 DPC4; BCL2; DCC; FVT1; PI5; SCCA1; SCCA2; GRP; MADH2; TGFBRE 19p p13.1 Amplification 2 (37%) JUND 19q q12-q13.1 Amplification 5 (50%) CCNE1; AXL; TGFB1; AKT2; CEBPA 19q q Gain 50 CCNE; AXL; TGFB1; AKT2; CEBPA; CEA; PRSSL1; BAX; FASB; PRSS9; PTPRH; BCL3 20q q Gain 68 GHRF; RBL1; E2F1; TOP1; AIB1; SRC; MYCL2; CAS; BCLX Xq q21-qter Loss 51 TDGF2; MYCL2; FGF13; MCF2; L1CAM a Genes are listed for the most frequent regions of loss and gain identified in this study (regions in which Ͼ50% of all tumors analyzed were altered). b Regions of amplification represent tumor:normal signal ratios in excess of 1.5; an additional 11 regions showed similar high levels of amplification but only in either one (1p32; 2q14.1-14.3; 6p21.3; 7q31; 10q22; 21q21-ter) or two (1q31; 3p13.3; 7p15-21; 15q22; 18q12-q21) tumors, respectively. c For amplifications, the figure in parentheses represents the frequency of gain in addition to amplification at that locus. d The candidate genes listed were obtained from the site http://cc.ucsf.edu/people/waldman/GENES/completechroms.html.

DISCUSSION to analyze 100 sporadic ovarian tumors identified multiple regions of loss and gain that are consistent with our findings (19). In this study, metaphase CGH analysis of 141 epithelial ovarian cancers identified multiple, frequently occurring somatic alterations. It is However, some of our data differ from previous studies (18–20); likely that many of these changes represent the location of genes that are for example, there are notable differences in the frequency with which critical for ovarian tumor development, although some may simply reflect alterations on chromosomes 3q, 5q, 6p, 12q, 17, 19p, 22q, and Xq an accumulation of genetic damage that occurs during tumor progression. were detected between this and other studies. Some of the disparity Some of the data from this study are consistent with previous between CGH and LOH data may be explained by differences be- metaphase CGH and LOH studies in ovarian cancer (18–20); for tween the two methods in their ability to resolve genetic alterations; example, regions of common deletion on chromosomes 4p, 6q, 9p, metaphase CGH paints a picture of gross genomic alterations, includ- 13q, 18q, and Xp have frequently been identified using LOH analysis ing changes in chromosome copy number, whereas LOH analysis (18). Similarly, a previous study in which metaphase CGH was used produces better, locus-specific resolution. However, LOH analysis is

Table 3 Comparison between the frequencies of somatic alteration identified at 20 different regions in BRCA1 and BRCA2-associated, familial non-BRCA1/2, and sporadic ovarian cancersa Pair-wise comparisons BRCA1 BRCA2 Non-1/2 Sporadic Region Loss/Gain (%) (%) (%) (%) ␹2 ● P abcdefghi j 1pter-p35 Gain 22 26 43 42 6.39 0.094 0.042 1cen-q21 Gain 22 57 27 36 9.22 0.026 0.013 0.032 0.006 3pter-p22 Gain 7 13 11 25 5.45 0.141 0.036 3cen-q13.3 Gain 13 16 7 25 6.61 0.086 0.04 6q24-qter Loss 54 52 55 29 5.84 0.120 0.02 8p12-cen Loss 26 26 34 11 4.94 0.176 0.029 9pter-p21 Gain 24 9 9 11 5.36 0.147 0.037 9cen-q13 Loss 15 22 21 46 10.00 0.019 0.005 0.006 0.034 11cen-q13 Gain 28 44 21 11 8.02 0.046 0.034 0.011 11q14-q22 Gain 15 13 7 0 5.53 0.137 0.04 12q21-23 Loss 33 13 25 11 6.22 0.101 0.049 12q21-23 Gain 4 22 5 21 10.04 0.018 0.047 0.049 0.042 0.037 13cen-q14 Loss 48 74 30 50 12.19 0.007 0.006 0.006 0.001 0.045 14q24-qter Gain 17 26 9 29 5.40 0.145 0.05 15q21-23 Gain 17 22 2 14 6.96 0.073 0.013 0.03 0.016 15q24-qter Gain 22 39 14 21 5.74 0.125 0.05 0.029 17p Loss 39 35 32 11 7.02 0.071 0.011 0.009 0.048 0.049 17cen-q21 Loss 41 13 27 14 9.36 0.025 0.014 0.02 0.027 18p Gain 4 30 16 21 8.99 0.029 0.012 0.047 0.005 22q12-qter Gain 13 8 14 29 4.77 0.190 0.044 a Pair-wise comparisons of the frequency of the change were carried out for the following 10 tumor group pairs: BRCA1 (a), BRCA2 (b), familial non-BRCA1/2 (c), and sporadic tumors (d) compared with the other tumor groups combined; sporadic tumors compared with BRCA1 (e), BRCA2 (f), and familial non-BRCA1/2 (g) tumor groups; familial non-BRCA1/2 tumors compared with BRCA1 (h) and BRCA2 (i) tumor groups; BRCA1 and BRCA2 tumor groups compared between each other (j). 420

Fig. 1. Hierarchical clustering of metaphase CGH data. Analysis was performed separately for BRCA1, BRCA2, familial non-BRCA1, and sporadic tumors. The presence of loss (red), gain (green), or no loss/gain (black) was recorded throughout the genome divided into 100 nonoverlapping regions of similar size. These data are displayed in a matrix format; each column within the matrix represents the spectrum of genetic alterations for a single tumor. For each matrix, the most densely grouped hierarchical cluster of loss and gain has been highlighted and the regions within the cluster described: red typeface refers to regions of loss that occur in clusters from three or more tumor groups; green typeface refers to regions of gain that occur in clusters from three or more tumor groups; black typeface refers to regions of loss or gain that occur in the clusters of only one or two tumor groups.

limited as a genome wide screen because it requires high-density influence the pattern of somatic genetic alterations during tumor microsatellite mapping, which is both time consuming to perform and development (21, 22). a considerably greater drain on DNA resources compared with CGH. We find evidence of a similar influence when the frequencies with Another reason for some of the differences between this and other which somatic genetic changes in tumors from BRCA1, BRCA2, metaphase CGH studies could be that whereas most previously pub- familial non-BRCA1/2, and sporadic cases are compared. We identi- lished CGH data are from sporadic ovarian cancers only, approxi- fied multiple differences between the four tumor groups, which sug- mately half of all tumors in this study were from BRCA1/2 mutation gests that they differ in some aspects of tumor development. However, carriers and only a fifth from sporadic cases; previous studies in breast we carried out a large number of significance tests, and it is likely that cancer suggest that the presence of germ-line BRCA1/2 mutation can some of these differences are chance occurrences. There were 200 421

Table 4 Somatic genetic events in ovarian tumors that are significantly associated with changes in disease stage and tumor grade Stage Grade

I/II III/IV 1 2 3 Region Alteration (n ϭ 17) (n ϭ 48) Pa (n ϭ 11) (n ϭ 44) (n ϭ 83) Pb P for trendc 1p32-p34.3 Gain 12% 40% 0.040 4q15.1-pter Loss 41% 8% 0.005 5q14-q22 Loss 47% 77% 0.032 40% 63% 73% 0.045 0.018 9p13-cen Loss 8% 22% 32% 0.093 0.044 11q14-q22 Loss 0% 14% 22% 0.047 0.056 21q22-qter Gain 0% 25% 0.027 22q12-qter Gain 39% 20% 10% 0.029 0.005 Xcen-q13 Loss 15% 17% 45% 0.002 0.002 a Fisher’s exact test. b ␹2 tests. c ␹2 tests for trend. individual comparisons between the four tumor groups and 32 alter- Peutz-Jeghers (24). More recently, CGH was used to identify a ations with a P Ͻ0.2 were selected for additional analyses in which an putative breast cancer susceptibility locus on chromosome 13q21-22 additional 320 pair-wise comparisons was performed. Applying the in a subset of predominantly Scandinavian non-BRCA1/2-associated Bonferroni correction to the results of these analyses would require a breast cancer families (25), although subsequent studies suggest that P of Ͻ0.00016 to achieve a conventional level of significance of if a susceptibility gene does exist at this locus, it is unlikely to be P Ͻ 0.05. The sample size of this study was not large enough to responsible for a substantial proportion of breast cancer families (26). generate such a small P, and indeed the smallest observed P was We have used metaphase CGH to identify somatic genetic alterations 0.001. However, we observed 41 pair-wise comparisons with signif- in ovarian tumors from families in which no BRCA1 or BRCA2 icant differences at the 0.05 level compared with 16 expected if there mutation could be identified. Somatic genetic alterations that are were no true differences in frequency of alteration between tumor frequent and/or specific to this group of tumors may represent the types, and 8 significant differences at the 0.01 level compared with 3 location of genes associated with inherited susceptibility to ovarian expected. This suggests that a substantial proportion of these differ- cancer. However, a degree of insensitivity in detecting BRCA1/2 ences are real. mutations and the possible existence of several rare high to moderate This assertion is supported by the observation, as expected, of and/or low penetrant ovarian cancer susceptibility genes (genetic significant increases in the frequencies of loss at the BRCA1 and heterogeneity) in non-BRCA1/2 families mean that we are unable to BRCA2 loci in tumors from BRCA1 and BRCA2 carriers, respec- suggest with confidence specific candidate susceptibility loci using tively. To provide a better indication of the most critical events, we the current data alone. used a hierarchical cluster algorithm to group alterations that tended to Some studies indicate that there are differences in the clinical occur together. The apparent clustering of a limited number of regions and/or histopathological characteristics of breast and ovarian tumors of loss and gain indicates a select series of targets for future studies between tumors from BRCA1, BRCA2, and non-BRCA1/2 mutation aimed at identifying genes in ovarian cancer. carriers (7–13). The reasons for this variation are unknown; BRCA1 Our findings in epithelial ovarian tumors are consistent with similar and BRCA2 may have a direct effect on the behavior of breast and studies in breast cancer (21, 22). In particular, the study of Tirkkonen ovarian epithelial cells, which could vary depending on germ-line et al. (21), which describes metaphase CGH analysis of 21 BRCA1, BRCA1/2 mutation status. Alternatively, BRCA1/2 mutation status 12 BRCA2, and 55 sporadic breast tumors, identified 17 significant might influence subsequent somatic genetic events in tumorigenesis (P Ͻ 0.05) differences in the frequencies of loss or gain between the with these events being responsible for the observed variation in three groups. As with this study, loss of the BRCA2 region was more tumor phenotype. The somatic genetic differences that we observed frequent in BRCA2 tumors than in BRCA1 and especially sporadic between BRCA1, BRCA2, non-BRCA1/2, and sporadic ovarian tu- tumors. However, in contrast to our findings, Tirkkonen et al. (21) mors provides support for the latter of these hypotheses. A more identified no losses of the BRCA1 region on chromosome 17q but did detailed comparison of the histopathological characteristics of ovarian detect frequent gain on this chromosome arm. tumors in BRCA1/2 mutation and nonmutation carriers will be re- The findings of a previous, similar study in epithelial ovarian quired to obtain a better understanding of this association. cancer (23) are in contrast to those of this study. Trapper et al. (23) In conclusion, we have used metaphase CGH to characterize the analyzed 16 BRCA1, 4 BRCA2, and 20 sporadic ovarian tumors and spectrum of somatic genetic events that occur in the development of found only one alteration, a gain on chromosome 2q24-q32, to be epithelial ovarian cancer in tumors from BRCA1 and BRCA2 muta- more frequent in BRCA1 tumors compared with non-BRCA1 tumors. tion carriers, familial non-BRCA1/2, and sporadic cases. In doing so, Our study also identified a higher frequency of gain at 2q24-q32 in we have identified molecular genetic differences between these four BRCA1 tumors compared with other tumor groups, although this did tumor groups that suggest there are different mechanisms for tumor not reach statistical significance. One explanation for the dissimilarity development, which may influence the phenotype and clinical out- in findings between the two studies could be that the limited statistical come of ovarian cancers. power resulting from the smaller number of tumors analyzed by Trapper et al. (23) precluded the identification of additional significant differences. ACKNOWLEDGMENTS Moderate or highly penetrant ovarian cancer susceptibility genes, The United Kingdom Coordinating Committee on Cancer Research Familial besides BRCA1 and BRCA2, that are responsible for familial clus- Ovarian Cancer Register is overseen by a steering committee, including the tering of ovarian cancer cases may exist. In the past, CGH analysis has following members: D. Timothy Bishop, William Collins, Douglas F. Easton, been successfully used for the localization and subsequent identifica- Gordon Fraser, Ian T. Jacobs, David Lowe, James Mackay, Bruce A. J. Ponder, tion of a gene associated with the cancer susceptibility syndrome John Shepherd, and C. Mike Steel. 422

REFERENCES Kerkhoven, R. M., Roberts, C., Linsley, P. S., Bernards, R., and Friend, S. H. Gene expression profiling predicts clinical outcome of breast cancer. Nature (Lond.), 415: 1. Miki, Y., Swensen, J., Shattuckeidens, D., Futreal, P. A., Harshman, K., Tavtigian, S., 530–536, 2002. Liu, Q. Y., Cochran, C., Bennett, L. M., Ding, W., Bell, R., Rosenthal, J., Hussey, C., 15. Piver, M. S., Baker, T. R., Jishi, M. F., Sandecki, A. M., Tsukada, Y., Natarajan, N., Tran, T., Mcclure, M., Frye, C., Hattier, T., Phelps, R., Haugenstrano, A., Katcher, H., Mettlin, C. J., and Blake, C. A. Familial ovarian cancer. A report of 658 families from Yakumo, K., Gholami, Z., Shaffer, D., Stone, S., Bayer, S., Wray, C., Bogden, R., the Gilda Radner Familial Ovarian Cancer Registry 1981–1991. Cancer (Phila.), 71: Dayananth, P., Ward, J., Tonin, P., Narod, S., Bristow, P. K., Norris, F. H., Helvering, 582–588, 1993. L., Morrison, P., Rosteck, P., Lai, M., Barrett, J. C., Lewis, C., Neuhausen, S., 16. Stratton, J., Gayther, S. A., Russell, P., Dearden, J., Gore, M., Blake, P., Easton, D., Cannonalbright, L., Goldgar, D., Wiseman, R., Kamb, A., and Skolnick, M. H. A and Ponder, B. A. J. The contribution of BRCA1 to ovarian cancer in a British strong candidate for the breast and ovarian-cancer susceptibility gene BRCA1. population. N. Engl. J. Med., 336: 1125–1130, 1997. Science (Wash. DC), 266: 66–71, 1994. 17. Eisen, M. B., Spellman, P. T., Brown, P. O., and Botstein, D. Cluster analysis and 2. Wooster, R., Bignell, G., Lancaster, J., Swift, S., Seal, S., Mangion, J., Collins, N., display of genome-wide expression patterns. Proc. Natl. Acad. Sci. USA, 95: 14863– Gregory, S., Gumbs, C., Micklem, G., Barfoot, R., Hamoudi, R., Patel, S., Rice, C., 14868, 1998. Biggs, P., Hashim, Y., Smith, A., Connor, F., Arason, A., Gudmundsson, J., Ficenec, 18. Jacobs, I., and Lancaster, J. The molecular genetics of sporadic and familial epithelial D., Kelsell, D., Ford, D., Tonin, P., Bishop, D. T., Spurr, N. K., Ponder, B. A. J., ovarian cancer. Int. J. Gynaecol. Cancer, 6: 337–355, 1996. Eeles, R., Peto, J., Devilee, P., Cornelisse, C., Lynch, H., Narod, S., Lenoir, G., 19. Kiechle, M., Jacobsen, A., Schwarz-Boeger, U., Hedderich, J., Pfisterer, J., and Egilsson, V., Barkadottir, R. B., Easton, D. F., Bentley, D. R., Futreal, P. A., Arnold, N. Comparative genomic hybridization detects genetic imbalances in primary Ashworth, A., and Stratton. M. R. Identification of the breast-cancer susceptibility ovarian carcinomas as correlated with grade of differentiation. Cancer (Phila.), 91: gene BRCA2. Nature (Lond.), 378: 789–792, 1995. 534–540, 2001. 3. Gayther, S. A., Russell, P., Harrington, P., Antoniou, A. C., Easton, D. F., and Ponder, 20. Zweemer, R. P., Ryan, A., Snijders, A. M., Hermsen, M. A. J. A., Meijer, G. A., B. A. J. The contribution of germline BRCA1 and BRCA2 mutations to familial Beller, U., Menko, F. H., Jacobs, I. J., Baak, J. P. A., Verheijen, R. H. M., Kenemans, ovarian cancer: no evidence for other ovarian cancer susceptibility genes. Am. J. P., and van Diest, P. J. Comparative genomic hybridization of microdissected familial Hum. Genet., 65: 1021–1029, 1999. ovarian carcinoma: two deleted regions on chromosome 15q not previously identified 4. Ford, D., Easton, D. F., Stratton, M., Narod, S., Goldgar, D., Devilee, P., Bishop, in sporadic ovarian carcinoma. Lab. Investig., 81: 1363–1370, 2001. D. T., Weber, B., Lenoir, G., ChangClaude, J., Sobol, H., Teare, M. D., Struewing, 21. Tirkkonen, M., Johannsson, O., Agnarsson, B. A., Olsson, H., Ingvarsson, S., Karhu, J., A., Scherneck, S., Peto, J., Rebbeck, T. R., Tonin, P., Neuhausen, S. R., Eyfjord, R., Tanner, M., Isola, J., Barkardottir, R. B., Borg, A., and Kallioniemi, O. P. Distinct J., Lynch, H., Ponder, B. A. J., Gayther, S. A., Birch, J. M., Lindblom, A., Stoppa- somatic genetic changes associated with tumour progression in carriers of BRCA1 Lyonnet, D., Bignon, Y., Borg, A., Hamann, U., Haites, N., Scott, R. J., Maugard, and BRCA2 germ-line mutations. Cancer Res., 57: 1222–1227, 1997. C. M., and Vasen, H. Genetic heterogeneity and penetrance analysis of the BRCA1 22. Hedenfalk, I., Duggan, D., Chen, Y. D., Radmacher, M., Bittner, M., Simon, R., and BRCA2 in breast cancer families. Am. J. Hum. Genet. 62:, 3: 676–689, 1998. Meltzer, P., Gusterson, B., Esteller, M., Kallioniemi, O. P., Wilfond, B., Borg, A., and 5. Smith, S. A., Easton, D. F., Evans, D. G. R., and Ponder, B. A. J. Allele losses in the region 17q12-21 in familial breast and ovarian-cancer involve the wild-type chromo- Trent, J. Gene-expression profiles in hereditary breast cancer. N. Engl. J. Med., 344: some Nat. Genet., 2: 128–131, 1992. 539–548, 2001. 6. Collins, N., McManus, R., Wooster, R., Mangion, J., Seal, S., Lakhani, S. R., 23. Tapper, J., Sarantaus, L., Vahteristo, P., Nevanlinna, H., Hemmer, S., Seppala, M., Ormiston, W., Daly, P. A., Ford, D., Easton, D. F., and Stratton, M. R. Consistent loss Knuutila, S., and Butzow, R. Genetic changes in inherited and sporadic ovarian of the wild-type allele in breast cancers from a family linked to the BRCA2 gene on carcinomas by comparative genomic hybridization: extensive similarity except for a chromosome 13q12-13. Oncogene, 10: 1673–1675, 1995. difference at chromosome 2q24–q32. Cancer Res., 58: 2715–2719, 1998. 7. Marcus, J. N., Watson, P., Page, D. L., Narod, S. A., Lenoir, G. M., Tonin, P., 24. Hemminki, A., Tomlinson, I., Markie, D., Jarvinen, H., Sistonen, P., Bjorkqvist, LinderStephenson, L., Salerno, G., Conway, T. A., and Lynch, H. T. Hereditary breast A. M., Knuutila, S., Salovaara, R., Bodmer, W., Shibata, de la Chapelle, A., and cancer, pathobiology, prognosis, and BRCA1 and BRCA2 gene linkage. Cancer Aaltonen, L. A. Localization of a susceptibility locus for Peutz-Jeghers syndrome to (Phila.), 77: 697–709, 1996. 19p using comparative genomic hybridization and targeted linkage analysis. Nat. 8. Breast Cancer Linkage Consortium. Pathology of familial breast cancer: differences Genet., 15: 87–90, 1997. between breast cancers in carriers of BRCA1 or BRCA2 mutations and sporadic 25. Kainu, T., Juo, S. H. H., Desper, R., Schaffer, A. A., Gillanders, E., Rozenblum, E., cases. Lancet, 349: 1505–1510, 1997. Freas-Lutz, D., Weaver, D., Stephan, D., Bailey-Wilson, J., Kallioniemi, O. P., 9. Johannsson, O. T., Idvall, I., Anderson, C., Borg, A., Barkardottir, R. B., Egilsson, V., Tirkkonen, M., Syrjakoski, K., Kuukasjarvi, T., Koivisto, P. R., Holli, K., Arason, A., and Olsson, H. Tumour biological features of BRCA1-induced breast and ovarian Johannesdottir, G., Bergthorsson, J. T., Johannsdottir, H., Egilsson, V., Barkardottir, cancer Eur. J. Cancer, 33: 362–371, 1997. R. B., Johannsson, O., Haraldsson, K., Sandberg, T., Holmberg, E., Gronberg, H., 10. Chang, J., Fryatt, I., Ponder, B., Fisher, C., and Gore, M. E. A matched control study Olsson, H., Borg, A., Vehmanen, P., Eerola, H., Heikkila, P., Pyrhonen, S., and of familial epithelial ovarian cancer: patient characteristics, response to chemotherapy Nevanlinna, H. Somatic deletions in hereditary breast cancers implicate 13q21 as a and outcome. Ann. Oncol., 6: 80–82, 1995. putative novel breast cancer susceptibility locus. Proc. Natl. Acad. Sci. USA, 97: 11. Rubin, S. C., Benjamin, I., Behbakht, K., Takahashi, H., Morgan, M. A., LiVolsi, 9603–9608, 2000. V. A., Berchuck, A., Muto, M. G., Garber, J. E., Weber, B. L., Lynch, H. T., and 26. Thompson, D., Szabo, C. I., Mangion, J., Oldenburg, R. A., Odefrey, F., Seal, S., Boyd, J. Clinical and pathological features of ovarian cancer in women with germ-line Barfoot, R., Kroeze-Jansema, K., Teare, D., Rahman, N., Renard, H., Mann, G., mutations of BRCA1. N. Engl. J. Med., 335: 1413–1416, 1996. Hopper, J. L., Buys, S. S., Andrulis, I. L., Senie, R., Daly, M. B., West, D., Ostrander, 12. Pharoah, P. D. P., Easton, D. F., Stockton, D. L., Gayther, S. A., and Ponder, B. A. J. E. A., Offit, K., Peretz, T., Osorio, A., Benitez, J., Nathanson, K. L., Sinilnikova, Survival in BRCA1 and BRCA2 associated epithelial ovarian cancer. Cancer Res., O. M., Olah, E., Bignon, Y. J., Ruiz, P., Badzioch, M. D., Vasen, H. F. A., Futreal, 59: 868–871, 1999. A. P., Phelan, C. M., Narod, S. A., Lynch, H. T., Ponder, B. A. J., Eeles, R. A., 13. Ramus, S. J., Fishman, A., Pharoah, P. D. P., Yarkoni, S., Alturas, M., and Ponder, Meijers-Heijboer, H., Stoppa-Lyonnet, D., Couch, F. J., Eccles, D. M., Evans, D. G., B. A. J. Ovarian cancer survival in Ashkenazi Jewish patients with BRCA1 and Chang-Claude, J., Lenoir, G., Weber, B. L., Devilee, P., Easton, D. F., Goldgar, D. E., BRCA2 mutations. Eur. J. Surg. Oncol., 27: 278–281, 2001. and Stratton, M. R. Evaluation of linkage of breast cancer to the putative BRCA3 14. van’t Veer, L. J., Dai, H. Y., van de Vijver, M. J., He, Y. D. D., Hart, A. A. M., Mao, locus on chromosome 13q21 in 128 multiple case families from the Breast Cancer M., Peterse, H. L., van der Kooy, K., Marton, M. J., Witteveen, A. T., Schreiber, G. J., Linkage Consortium. Proc Natl Acad Sci USA, 99: 827–831, 2002.

423

Downloaded from cancerres.aacrjournals.org on September 25, 2021. © 2003 American Association for Cancer Research. BRCA1/2 Mutation Status Influences Somatic Genetic Progression in Inherited and Sporadic Epithelial Ovarian Cancer Cases

Susan J. Ramus, Paul D. P. Pharoah, Patricia Harrington, et al.

Cancer Res 2003;63:417-423.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/63/2/417

Cited articles This article cites 25 articles, 8 of which you can access for free at: http://cancerres.aacrjournals.org/content/63/2/417.full#ref-list-1

Citing articles This article has been cited by 2 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/63/2/417.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/63/2/417. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.