Copy Number Aberrations in Benign Serous Ovarian Tumors: a Case for Reclassiﬁcation?

Published OnlineFirst October 5, 2011; DOI: 10.1158/1078-0432.CCR-11-2080

Clinical Cancer Human Cancer Biology Research

Copy Number Aberrations in Benign Serous Ovarian Tumors: A Case for Reclassiﬁcation?

Sally M. Hunter1, Michael S. Anglesio6, Raghwa Sharma3, C. Blake Gilks6,7, Nataliya Melnyk6, Yoke-Eng Chiew4,5, Anna deFazio4,5 for the Australian Ovarian Cancer Study Group, Teri A. Longacre8, David G. Huntsman6,7, Kylie L. Gorringe1,2, and Ian G. Campbell1,2

Abstract Purpose: Serous ovarian carcinomas are the predominant epithelial ovarian cancer subtype and it has been widely believed that some or all of these may arise from precursors derived from the ovarian surface epithelium or fimbriae, although direct molecular evidence for this is limited. This study aimed to conduct copy number (CN) analysis using a series of benign and borderline serous ovarian tumors to identify underlying genomic changes that may be indicative of early events in tumorigenesis. Experimental Design: High resolution CN analysis was conducted on DNA from the epithelial and fibroblast components of a cohort of benign (N ¼ 39) and borderline (N ¼ 24) serous tumors using the Affymetrix OncoScan assay and SNP6.0 arrays. Results: CN aberrations were detected in the epithelium of only 2.9% (1 of 35) of serous cystadenomas and cystadenofibromas. In contrast, CN aberrations were detected in the epithelium of 67% (16 of 24) of the serous borderline tumors (SBT). Unexpectedly, CN aberrations were detected in the fibroblasts of 33% (13 of 39) of the benign serous tumors and in 15% (3 of 20) of the SBTs. Of the 16 cases with CN aberrations in the fibroblasts, 12 of these carried a gain of chromosome 12. Conclusions: Chromosome 12 trisomy has been previously identified in pure fibromas, supporting the concept that a significant proportion of benign serous tumors are in fact primary fibromas with an associated cystic mass. This is the first high resolution genomic analysis of benign serous ovarian tumors and has shown not only that the majority of benign serous tumors have no genetic evidence of epithelial neoplasia but that a significant proportion may be more accurately classified as primary fibromas. Clin Cancer Res; 17(23); 7273–82. 2011 AACR.

Introduction At the time of diagnosis, women with epithelial ovarian cancer usually have advanced disease and as a consequence Ovarian cancer is a very significant health burden and the their prognosis is extremely poor (5-year survival rate for 7th leading cause of cancer death in women worldwide (1). stage III and IV disease is only 25%–30%; refs. 2, 3). For such a clinically significant disease, remarkably little is known about the molecular events that initiate the disease. Authors' Affiliations: 1Centre for Cancer Genomics and Predictive Med- Although the paradigm that malignancies arise through a icine, Peter MacCallum Cancer Centre, Melbourne; 2Department of Pathol- ogy, University of Melbourne, Parkville, Victoria; 3Anatomical Pathology, stepwise progression from benign precursors has been University of Sydney and University of Western Sydney at Westmead established for many malignancies, the archetypical exam- 4 5 Hospital; Department of Gynaecological Oncology, Westmead Institute ple being colorectal carcinogenesis (4), it remains unclear if for Cancer Research,University of Sydney at Westmead Millennium Institute, Westmead Hospital, Sydney, New South Wales, Australia; 6The Department this holds true for ovarian cancer. There is still considerable of Pathology and Laboratory Medicine, University of British Columbia; controversy as to what constitutes a true ovarian cancer 7 Genetic Pathology Evaluation Centre of the Prostate Research Centre and precursor; an important definition that needs to be made to Department of Pathology, Vancouver General Hospital and University of British Columbia, Vancouver, British Columbia, Canada; 8Department of understand the origins and identify new clinical interven- Pathology, Stanford University School of Medicine, Stanford, California tions for this lethal disease. Note: Supplementary data for this article are available at Clinical Cancer Serous ovarian carcinomas are the predominant clinically Research Online (http://clincancerres.aacrjournals.org/). important subtype but at present there is little experimental

K.L. Gorringe and I.G. Campbell are co-senior authors. evidence from which to draw convincing conclusions about what constitutes the precursor(s) to this subtype. It has been Corresponding Author: Ian G. Campbell, VBCRC Cancer Genetics Research Laboratory, Peter MacCallum Cancer Centre, Locked Bag 1, widely believed that some or all of these arise from pre- A'Beckett Street, Melbourne, Victoria 8006, Australia. Phone: 613-9656- cursors originating from the ovarian surface epithelium, 1803; Fax: 613-9656-1411; E-mail: [email protected] via inclusion cysts or serous benign and borderline tumors doi: 10.1158/1078-0432.CCR-11-2080 (5–8). Obvious candidate precursors are serous ovarian 2011 American Association for Cancer Research. cystadenomas and cystadenoﬁbromas, which are benign

www.aacrjournals.org 7273

Hunter et al.

cursors to some invasive serous ovarian carcinomas, we Translational Relevance have conducted high-resolution CN analysis on a series of benign and borderline serous ovarian tumors. This is the Ovarian cancer is a very significant health burden and first ultra-high-resolution CN analysis of benign serous the seventh leading cause of cancer death in women. At tumors of the ovary. the time of diagnosis, women often have advanced disease and as a consequence their prognosis is extremely Materials and Methods poor. Our understanding of the progression of ovarian cancer through precursor stages and the molecular genet- Tissue samples ic events underlying these changes is very limited. Only fresh frozen tissue samples were used in this for CN Although a number of candidate precursor lesions have and mutation analyses. All samples were collected with the been proposed, the true contribution of these precursor informed consent of patients and the study was approved by lesions to the onset of ovarian cancer is unresolved. the Human Research Ethics Committees at the Peter Mac- Identifying genuine ovarian cancer precursors and defin- Callum Cancer Centre, Queensland Institute of Medical ing key molecular genetic events initiating and promot- Research, University of Melbourne and all participating ing tumorigenesis have important implications in early hospitals. Patients with ovarian tumors were identified detection and treatment of ovarian cancers. through 2 primary sources: (i) 9 at hospitals in South- ampton (24), United Kingdom, (ii) 54 through the Austra- lian Ovarian Cancer Study (25). Pathology reviews were done independently by 2 gynecologic pathologists (R. lesions with a cystic mass of 1 cm or more in diameter, lined Sharma and C.B. Gilks). Pathology review was conducted with a single layer of cuboidal to columnar epithelium and on cryosections adjacent to the tissue from which DNA was commonly associated with a fibromatous stromal mass extracted (n ¼ 63). (Supplementary Fig. S1). The serous epithelial layer of these tumors typically displays minimal cellular proliferation and Microdissection and DNA extraction no nuclear atypia. These are relatively common tumors, A representative hematoxylin and eosin (H&E) stained accounting for 60% of all serous ovarian tumors, whereas section was assessed and needle microdissection was done serous borderline tumors (SBT) and low-grade serous car- using 10 mm sections to obtain high percentage tumor cinomas (LGSC) account for 10%–15% and 2%–9% of all epithelial cell and fibroblast cell components. DNA was serous ovarian tumors, respectively (9–12). Benign serous extracted using the Qiagen Blood and Tissue Kit (Qiagen). tumors are presumed by many to be precursors to SBTs Normal DNA extracted from blood lymphocytes was avail- based on similarities in the cystic structure of some SBTs and able for all 63 patients. the frequent detection of cases with a benign cyst coexisting with a SBT or cases comprising predominantly benign cysts CN arrays with regions of atypical proliferation (13). Despite the A subset of cases were processed by Affymetrix for the cooccurrence of benign, borderline, and low-grade carci- OncoScan (Molecular Inversion Probe) assay, which con- noma epithelial components, direct molecular evidence sists of a 330K probe set that allows the detection of supporting benign lesions as precursors is limited. Although genome-wide, allele-specific CN. OncoScan data normal- some studies have shown the existence of KRAS and BRAF ization was done by Affymetrix Inc. as previously mutations in ovarian serous cystadenomas and cystadeno- described (26, 27). Where sufficient material ( 250 ng fibromas co-existing with a region of atypical proliferation DNA) was available, the Affymetrix SNP6.0 (1.8M probe or adjacent SBT (14), mutations in these genes have not set) arrays were utilized for ultra-high-resolution allele- been detectable in solitary benign tumors. specific CN analysis, although prior to its release the SBTs have been firmly established as the likely precursor Affymetrix 500K array was used. For the SNP6.0 array lesions to LGSCs, sharing similar rates of KRAS and BRAF the input was reduced from the recommended 500 to 250 mutation and low levels of genomic instability (15–18). ng with no detectable difference in the quality of the data. This is in contrast to the high rates of TP53 mutation and Reaction volumes were halved accordingly prior to the high levels of genomic alteration observed in high-grade SNP6.0 PCR step. MAPD scores (pass 0.4) for samples serous carcinomas (18, 19). Traditionally, these have been run on OnsoScan and SNP6 platforms are available in believed to arise from ovarian surface epithelium and Supplementary Tables S5 and S6. epithelial inclusion cysts formed from invaginated surface epithelium (5). More recently, circumstantial evidence has Data analysis been published suggesting that serous ovarian carcinomas The SNP and OncoScan data were analyzed using Partek may not arise from the ovary at all and may in fact originate Genomics Suite 6.5, using paired and unpaired CN gener- from fallopian tube epithelium (20, 21, 22, 23). ation, allele-specific CN analysis and circular binary seg- With the aim of identifying somatic genomic changes that mentation (CBS) to identify regions of CN aberration and may be indicative of early events in tumorigenesis and LOH. Regions of CN aberration and LOH were confirmed which could assist in determining if these represent pre- through examination of allele-specific CN ratios.

7274 Clin Cancer Res; 17(23) December 1, 2011 Clinical Cancer Research

Copy Number Aberrations in Benign Serous Ovarian Tumors

Table 1. CN analysis of benign serous tumors

Sample ID Morphology Epithelium CN Fibroblast CN Affymetrix aberrations aberrations platform

5 Serous cystadenofibroma Gain: 6p, 7q; LOH: 6q, Gain: 12; LOH: 22 OncoScan 7p, 13q12.11-12.12 467 Serous cystadenofibroma None Gain: 9q, 16q12.1-12.2; OncoScan LOH: 16q13-24.3 A1 Serous cystadenofibroma LOH: Xa LOH: X (low level)a OncoScan A2 Serous cystadenofibroma None Gain: 12 OncoScan A3 Serous cystadenofibromac None Gain: 12 OncoScan A4 Serous cystadenofibromac None Gain: 12; LOH: 17q, 22 OncoScan A5 Serous cystadenofibroma None Gain: 12 OncoScan 158 Serous cystadenofibroma None Gain: 12 SNP6.0 450 Serous cystadenofibroma None LOH: 22 SNP6.0 A6 Serous cystadenofibroma None Gain: 12 SNP6.0 A7 Serous cystadenofibroma None Gain: 12 SNP6.0 A8 Serous cystadenofibroma None None OncoScan A9 Serous cystadenofibroma None None OncoScan A10 Serous cystadenofibromac None None OncoScan A11 Serous cystadenofibroma None None OncoScan A12 Serous cystadenofibroma None None OncoScan A13 Serous cystadenofibroma None None OncoScan A14 Serous cystadenofibroma None None OncoScan 103 Serous cystadenofibroma None None 500K/SNP6.0 164 Serous cystadenofibroma None None SNP6.0 A15 Serous cystadenofibroma None None SNP6.0 A16 Serous cystadenoma None LOH: 3p21.33-14.3, OncoScan 7q11.21-11.23, 7q22.1 A17 Serous cystadenomab None Gain: 8, 10, 12, 13, 15, 18, 19 OncoScan A18 Serous cystadenoma None None OncoScan A19 Serous cystadenoma None None OncoScan A20 Serous cystadenoma None None OncoScan A21 Serous cystadenoma None None OncoScan A22 Serous cystadenomab None None OncoScan 7 Serous cystadenomab None None 500K/SNP6.0 148 Serous cystadenoma None None 500K/SNP6.0 A23 Serous cystadenoma None None SNP6.0 A24 Serous cystadenomab None None SNP6.0 A25 Serous cystadenoma, papillary None None OncoScan A26 Serous cystadenoma, papillary None None SNP6.0 A27 Serous cystadenoma, papillaryb None None OncoScan

aApparent germ line mosaic loss of X with loss of same allele in all tissues, therefore not considered a tumor-speciﬁc loss. bFibrous stroma noted on review. cFibrous stroma not noted on review.

Mutation screening FISH and aneusomy DNA sequencing was done by Sanger sequencing using FISH was carried out on 6 mm formalin-fixed, paraffin- BDT v3.1 reagents (Applied Biosystems) and an ABI3130 embedded (FFPE) sections using the following commercial sequencer. Sequencing was used to identify the most com- and in-house BAC probes: Cep12 (orange, 12p11.1-q11, mon serous ovarian tumor mutations: BRAF codon 600, Abbottt Molecular 30-160012), RP11-1K3 Green d-UTP KRAS codons 12 and 13, TP53 exons 5–8 and ERBB2 exon (chr12q23.3-q24.11), RP11-13G14 Green d-UTP 20 (Supplementary Fig. S3). Primer sequences are detailed (chr12q23.3-q24.11), and RP11-209C18 Alexa Fluor 647 in Supplementary Table S1 and specific mutations identi- d-UTP (blue, chr15q25.3). Bacteria artificial chromosome fied are detailed in Supplementary Table S2. probes were labeled using nick translation kit (Abbott

www.aacrjournals.org Clin Cancer Res; 17(23) December 1, 2011 7275

Hunter et al.

Figure 1. Whole genome CN plot. Benign serous cystadenofibroma (case 5) identified with unique CN aberrations in the epithelium (þ6p, þ7q, -6q, -7p) and adjacent fibroblasts (þ12, -22).

Chr 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 x

Molecular; #32-801300). Aneusomy analysis was con- (75%) showed gain of chromosome 12 and 3 (25%) ducted on 6 mm FFPE sections using a commercial assay showed LOH of chromosome 22. from Abbott Molecular (formerly Vysis), Breast Cancer In a further attempt to identify evidence of somatic Aneusomy Multi-Color Probe Kit: Vysis LSI 1 - 1p12 Spec- genetic events in the epithelial component, each sample trum Gold (yellow), Vysis CEP 8 -8p11.1 Alpha Satellite was analyzed for mutations in KRAS codons 12 and 13, DNA Spectrum Red, Vysis CEP 11 - 11p11.11-q11 Spectrum BRAF codon 600, TP53 exons 5–8 and ERBB2 exon 20 but Green, and Vysis CEP 17 - 17p11.1-q11.1 Spectrum Aqua. no mutations were detected in the either the fibroblast or All hybridizations were carried out as previously described epithelial component of any case. (28). Imaging was done on a Zeiss Axioplan epifluorescent The observed rate of CN/LOH alterations in the fibro- microscope. blasts was nearly 3 times greater in cystadenofibromas compared with cystadenomas. However, as these cases were Results selected on the basis of the presence of microdissectible epithelium we considered that it was possible they may not Benign serous tumor CN analysis have been representative of the typical spectrum of benign H&E-stained sections from a cohort of benign serous serous tumors. Therefore, the fibroblast component from tumors (serous cystadenomas and cystadenofibromas) 10 consecutive benign serous tumors [5 cystadenomas and 5 were reviewed to identify cases where sufficient epithelial cystadenofibromas; 6 of these are also reported in (Table 1)] and stromal tissue could be microdissected to conduct high- were analyzed for CN/LOH alterations using SNP6.0 arrays resolution, genome-wide, allele-specific CN analysis. A total (Supplementary Table S4). The frequency of CN/LOH of 14 serous cystadenomas and 21 serous cystadenofibro- alterations present in the fibroblasts from these cystadenomas were identified where CN could be analyzed in match- fibromas was similar to the original set of selected cases ing epithelium, fibroblasts, and germ line (lymphocyte) (40.0% vs. 42.9%) but was much higher among the cysta- DNA. denomas (40% vs. 14.3%) although this was not statisti- All tissues were microdisssected to achieve more than cally significant (P ¼ 0.2722, the Fisher exact test). All 4 80% pure epithelial or fibroblast cell populations (Supple- cases with CN/LOH alterations included gain of chromo- mentary Fig. S2), which we have shown previously is some 12. Among the entire cohort, 13 of 39 cases (33.3%) sufficient purity to reliably identify CN aberrations and showed CN/LOH alterations in the fibroblast component. LOH (29). CN aberrations and/or LOH were detected in the epithelial component in one of the 35 cases (2.9%) but SBT CN analysis surprisingly, 12 of the 35 cases (34.3%) harbored CN Given that many of the benign serous tumors seemed to aberrations and/or LOH in the fibroblast component be neoplastic fibromatous masses rather than genuine epi- (Table 1). In one of these cases unique CN changes were thelial tumors it was plausible that a similar situation might detected in both the epithelium and fibroblasts, consistent also exist for SBTs. Consequently we conducted high-reso- with this case comprising 2 distinct and independent lution, genome-wide CN analysis on epithelial and fibro- tumors (Fig. 1). Overall, fibroblast components from blast components from a series of SBTs. Among the 24 SBTs 14.3% of the serous cystadenomas and 47.6% of the cysta- analyzed, CN/LOH aberrations were detected in 16 of 24 denofibromas harbored CN/LOH alterations (Supplemen- epithelial components (67%) and in 3 of 20 fibroblast tary Table S3). Among the 12 cases with CN/LOH changes, 9 components (15%; Table 2). BRAF or KRAS mutations were

7276 Clin Cancer Res; 17(23) December 1, 2011 Clinical Cancer Research

Copy Number Aberrations in Benign Serous Ovarian Tumors

Table 2. CN aberrations borderline serous tumors

Sample ID Morphology Epithelium CN aberrations Fibroblast CN Affymetrix Mutations aberrations platform

A32 SBT, NOS Gain: 8q21.3-q22.1, 8q22.2-qter; None SNP6.0 BRAF V600E CN LOH: 7q A33 SBT, NOS Gain: 7, X None SNP6.0 BRAF V600E CN LOH: 7q22.1-qter A34 SBT, NOS CN LOH: 7q11.22-qter Untested SNP6.0 BRAF V600E A35 SBT, NOSa LOH: X None SNP6.0 KRAS G12V A36 SBT, NOS Gain: 2, 7, 8, 12, 18 None SNP6.0 KRAS G12D A37 SBT, NOSa LOH: 1p35.1-36.32, 19 Untested SNP6.0 KRAS G12V A38 SBT, adenoﬁbromaa Gain: 8q, 12p None SNP6.0 KRAS G12V A39 SBT, NOSa Gain: 2, 7, 8, 12, 18, 20 None SNP6.0 WT A40 SBT, NOS CN LOH: 12p None SNP6.0 WT A41 SBT, NOSa Gain: 8q; None SNP6.0 WT LOH: 1p35.3-pter A42 SBT, papillary cystica CN LOH: 17q None SNP6.0 WT A43 SBT, papillary CN LOH: 19q13.32-13.43 None SNP6.0 WT A44 SBT, papillarya Gain: 1q; LOH: 16q None SNP6.0 WT A45 SBT, NOS Gain: 7q32.2; Gain: 12 SNP6.0 WT CN LOH: 11q13.5-qter A46 SBT, NOSa CN LOH: 17q Gain: 12, 14, 22q11.21; SNP6.0 WT LOH: 21 A47 SBT, cystica Gain: 8q22.1-qter; LOH: X SNP6.0 WT LOH: 6q23.2-qter 558 SBT, NOS None None SNP6.0 BRAF V600E A48 SBT, NOS None None SNP6.0 BRAF V600E A49 SBT, adenoﬁbromaa None None SNP6.0 BRAF V600E A50 SBT, papillary None None SNP6.0 BRAF V600E A51 SBT, papillary cystica None None SNP6.0 BRAF V600E A52 SBT, papillary cystic None Untested SNP6.0 BRAF V600E A53 SBT, papillary cystica None Untested SNP6.0 BRAF V600E A54 SBT, papillary cystic None None SNP6.0 WT

Abbreviations: NOS, not otherwise speciﬁed; WT, wild-type. aFibrous stroma noted on review.

detected in the epithelial component in 57.9% of cases mosome 12 in the majority of fibroblasts. Interestingly, A2 (Table 2); however, no TP53 or ERRB2 mutations were seemed to have 4 copies of chromosomes 12 and 15 (a detected. Overall, CN/LOH and/or BRAF/KRAS mutations chromosome 15 probe was used as a control) in some of were detected in the epithelial component of all but one of the epithelial cells (Fig. 2). Balanced tetrasomy in clusters the borderline tumors. In all 3 cases where CN aberrations of epithelial cells was confirmed using an aneusomy were detected in the fibroblasts component, unique CN detection kit and centromeric probes for chromosomes aberrations were also detected in the adjacent epithelial 1, 8, 11, and 17 (Supplementary Fig. S5). Completely component showing they were not clonal variants of a balanced tetrasomy is not readily detectable by SNP array common initiating tumor (Table 2, Supplementary Fig. S4). as CN is calculated after a median-centered normalization procedure. Further examination of the SNP allelic ratio FISH was also unable to detect imbalance, suggesting either this As noted above, gain of chromosome 12 was the most duplication is nearly perfectly balanced or an insufficient common CN aberration among stromal components of proportion of affected cells were present to make the benign tumors. To confirm the presence of the chromosome distinction. FISH was also carried out on a tumor micro- 12 gain in the fibroblasts from the benign serous tumors array with tissues from normal ovary (6), benign cord- and to determine absolute CN, FISH was carried out on 2 stromal tumors (3), benign endometriosis (4), and SBTs cases where suitable tissue was available (cases A2 and A6). (42). Only a single SBT was identified with chromosome Both A2 and A6 were confirmed to have trisomy of chro- 12 trisomy in approximately 45% of fibroblast cells,

www.aacrjournals.org Clin Cancer Res; 17(23) December 1, 2011 7277

Hunter et al.

Figure 2. Triploidy and tetrasomy. Case A2 was found to have clusters of epithelial cells with balanced tetrasomy (arrow) alongside ﬁbroblast cells triploid for chromosome 12 (arrowheads). Probes: orange, 12p11.1-q11; green, 12q23.3-q24.11; blue, 15q25.3.

although this case also had features of low-grade serous of KRAS and BRAF mutations in ovarian serous cystadeno- carcinoma/psammocarcinoma. mas and cystadenofibromas have been limited to a small number of cases with a coexisting region of atypical prolif- Discussion eration or adjacent SBT (14). Despite the small number of CN/LOH aberrations present in the epithelium of benign This study has shown that CN/LOH aberrations in the serous tumors, combining our data with published studies epithelium of benign serous tumors are rare, occurring in shows that they have similar profiles to a proportion of just one (2.9%) of cases studied, whereas aberrations in the SBTs, such as gain of 6p and 7q. If the benign serous tumors fibromatous components were more frequent, occurring in with CN/LOH aberrations are indeed precursors to SBT, the 33.3% of cases. Where CN/LOH aberrations were observed absence of either KRAS or BRAF mutations, characteristic of in the epithelial components these were similar to those SBTs, might indicate that acquisition of these mutations identified in previous studies of benign serous tumors accompanies progression to a SBT. This possibility is con- (Table 3; ref. 30, 31). Although the rate of epithelial CN/ sistent with the observations of Singer and colleagues (ref. 6) LOH events in this study is significantly lower than previ- who showed that bilateral SBTs frequently shared a subset of ously reported, artifactual LOH events may have been CN aberrations but rarely shared identical KRAS mutations, introduced when conducting microsatellite analyses on low indicating that the CN aberrations are likely occurring quantities of DNA in these previous studies (32). Our earlier than the acquisition of KRAS mutations and precede findings are consistent with the findings of Cheng and dissemination to the contralateral ovary. CN aberrations as colleagues (33) who showed in a cohort of 29 ovarian early events in ovarian serous neoplasia is further supported serous cystadenomas that only 14% of these tumors had by studies that have identified aneuploidy, but not muta- detectable monoclonal expansion in the epithelium. tions, in ovarian inclusion cysts and even at low levels in The absence of detectable KRAS, BRAF, ERBB2,orTP53 ovarian surface epithelium (7, 8). Alternatively, the absence mutations in the epithelial component of any benign serous of identifiable KRAS or BRAF mutations in benign serous lesion in this study is consistent with the low rate of such tumors with clear clonal expansion events leaves open the mutations reported from previous studies (17, 31). Reports possibility that they may be precursors for other serous

7278 Clin Cancer Res; 17(23) December 1, 2011 Clinical Cancer Research

Copy Number Aberrations in Benign Serous Ovarian Tumors

Table 3. Summary of serous CN/LOH studies

Benign CN aberrations SBT CN aberrations Reported benign Reported SBT Reported LGSC CN (this study) (this study) CN aberrations CN aberrations aberrations

Gain Gain Gain Gain 6p, 7q 1q, 2, 7, 8, 8q, 1q, 6p 1q, 2, 2q, 6p, 6q, 8q21.3-q22.1, 7, 8, 8q, 9p, 12, 8q22.1-qter, 10, 12, 13q, 16, 16p 12p, 18,20 LOH LOH LOH LOH LOH 6q, 7p, 13q12.11-12, X. 1p35.1-pter, 6q23.3, 1p32-p11, 4q13-34, 1p, 1p36,4, 9, 1p36, 9p21.3 7q, 11q13.5-qter, 5q11-q23, 9p21.3, 12q, 12p, 16q, 17q, 6q12-q23, 6q16.3, 14q, 15q, 16p, 19q13.32-13.43, 6q22.2, 7p22.2, 17, 17p, 19, X 7p15.3, 7p12.3, 17q, 19p, 19q, 22q 7q22.1, 7q31.1, 7q36.1, 9p21.3, 11q23.3 AI AI 1p, 5q, 8p, 18q, 1p, 5q, 8p, 18q, 22q, Xp 22q, Xp

NOTE: Benign (30, 31, 51); SBTs (18, 51, 52); LGSCs (6, 18, 51). Bold indicates overlapping CN aberrations between current study and previous reports.

ovarian cancers. The tetrasomy identified in the epithelial inclusions. Consequently, they concluded that the majority component in case A2, adjacent to fibroblasts with chro- of these tumors should not be considered serous neo- mosome 12 trisomy, is a particularly intriguing finding in plasms. Consistent with finding of the study of Seidman light of ploidy studies by Pradhan and colleagues that and Mehrota of greater evidence of "true" neoplasia in identified tetraploidy in 7 of 40 high-grade serous carcino- cystadenofibromas compared with cystadenomas, we only mas, 0 of 22 LGSCs, and 0 of 245 SBTs (34). observed CN aberrations in the epithelium of cystadenofi- In contrast to the benign serous tumors, CN/LOH events bromas and none in the epithelium of cystadenomas. were identified in the epithelial component in 16 of 24 Pathologist reviews of the cases in this study failed to (67%) of SBTs; similar aberrations have been detected in identify any distinguishing histologic features associated previous studies (Table 3). Although limited CN/LOH data with detectable CN aberrations in the fibromatous compo- is available that distinguishes low-grade serous from high nents compared with those tumors without detectable grade serous and other histologic types of ovarian cancer, fibromatous CN aberrations. One explanation may be that overlapping CN/LOH events support the model of SBTs as in fact the majority of these benign tumors do contain precursors to LGSCs (Table 3). Among the borderline clonal fibromatous neoplasms that this molecular study tumors 57.9% harbored KRAS or BRAF mutations, consis- has not detected due to the focus on CN analysis and tent with previous reports for both SBTs and LGSCs mutations most commonly present in the epithelium of (18, 35). serous ovarian tumors. To date no highly recurrent mutated An important finding of this study is that the fibromatous genes have been identified in fibromas that can be used to components of benign tumors frequently harbor identifi- assess the true frequency of benign serous tumors that are able clonal CN/LOH alterations, strongly suggesting that fibromas. many tumors originally classified as benign epithelial Trisomy 12 has been previously identified as a charac- tumors are in fact primary fibroblastic tumors with an teristic CN aberration of the fibroma-thecoma group of sex associated epithelial cyst. This supports the work of Seid- cord-stromal tumors (37–40). Persons and colleagues man and Mehrota (ref. 36), who published a review of 113 ref. 41) found chromosome 12 gain in 40% of pure fibro- unselected benign serous ovarian tumors to assess the rate mas, very similar to the rate detected in the fibroblasts of this of epithelial "neoplasia" based on the presence of a min- study. Although chromosome 12 gain has previously been imum of 1 mm (2) area of epithelial proliferation, finding identified in benign ovarian tumors these studies were done that only 7% of cases passed this criterion. Seidman and on whole tumors and therefore failed to identify the cell Mehrotra predicted that 81% of serous cystadenofibromas type of origin (40). Our data clearly show for the first time are in reality fibromas with epithelial inclusions and that that at least 40% of benign serous tumors are likely primary 99% of serous cystadenomas are cystically dilated glandular fibromas. It has been previously speculated that recurrent

www.aacrjournals.org Clin Cancer Res; 17(23) December 1, 2011 7279

Hunter et al.

Normal epithelial cell Genetically altered epithelial cell

Normal fibroblast cell

Genetically altered fibroblast cell

Fibroblast secreted factor Epithelial secreted factor B

Figure 3. Primary fibroblastic neoplasia. A, genetically altered fibroblast secretes growth/ proliferation signals to influence surrounding cells. B, clonal expansion of genetically altered fibroblasts with secretion of growth/ proliferation signals that influence surrounding cells and adjacent epithelium to form benign glandular structures. C, genetic alteration of epithelial cells resulting in neoplasia in response to adjacent fibroma. C

gain of chromosome 12 may be driven by a requirement for high-grade endometrioid tumor. Although this was a single an extra copy of the KRAS oncogene for neoplasia in this case in 25 cases analyzed, the presence of an underlying milieu (42). A correlation between trisomy 12 and fibroma in another histologic subtype suggests this phe- increased KRAS expression has been shown in an adenos- nomenon may not be limited to serous tumors. This finding quamous carcinoma of the lung (43), however, this has not raises some interesting questions about the potential bio- been shown experimentally in ovarian tumors. Other logical relationship between these tumor populations: are known oncogenes are located on chromosome 12, includ- these biphasic tumors, collision tumors, or is it possible that ing CDK4, CCND2, MDM, WNT1, and ERBB3, along with a the fibromatous component promotes tumorigenesis in the number of other genes of potential relevance in neoplasia. adjacent epithelium? The absence of shared point muta- An intriguing finding of this study was the identification tions and CN aberrations (within the limits of our assays) of one benign tumor and 3 borderline tumors with coex- between the adjacent epithelial and fibroblast components isting but unique CN aberrations in the epithelial and would seem to rule out a mesenchymal–epithelial or epi- fibroblast components. Interestingly, loss of chromosome thelial–mesenchymal transition event underlying a biphas- 22, the second most frequent CN aberration in fibromas ic tumor. The likelihood of these being collision tumors also detected in this study, was also detected by Qiu and col- seems highly unlikely given that the frequency of fibromas leagues (2005; ref. 29) in the fibroblast component of a and genuine benign serous tumors is extremely low (44).

7280 Clin Cancer Res; 17(23) December 1, 2011 Clinical Cancer Research

Copy Number Aberrations in Benign Serous Ovarian Tumors

Speculatively, our data points to a unique ovarian tumor- This is the largest high-resolution study of benign serous igenesis pathway whereby a pre-existing fibroma promotes tumors conducted to date and provides evidence that the proliferation and occasional transformation of the adjacent majority of benign ovarian serous tumors are not epithelial ovarian epithelium. Fibroblasts have long been recognized neoplasms, but are in fact primary fibromas. This finding as an integral part of the tumor microenvironment influ- significantly deflates the prevalence of serous histologic encing tumorigenesis, capable of both inhibitory and stim- subtype of ovarian neoplasms, of which benign tumors ulatory influences on neoplasia in neighboring epithelia accounted for 60%. The findings of this study also provide (45). In model systems, artificially induced genetically preliminary evidence for a possible novel pathway of altered fibroblasts have been previously shown to promote tumorigenesis dependent on a promoting primary fibroma. both initiation and progression of tumorigenesis in adjacent epithelia (46–48). It has been proposed that in primary Disclosure of Potential Conflicts of Interest human cancers, adjacent fibroblasts (so called cancer associated fibroblasts) almost universally acquired clonal No potential conflicts of interest were disclosed. somatic genetic mutations in genes that promoted the progression of the adjacent cancer epithelium. Although Acknowledgments that theory has subsequently been disproved (29), our data for the first time suggests that in rare instances a genuine The authors thank the cooperation of the participating institutions in Australia; the contribution of the study nurses, research assistants, and all stromal tumor can promote tumorigenesis in adjacent clinical and scientific collaborators; and all of the women who participated in epithelium. A possible mechanism for this to occur could the study. Members of the Australian Ovarian Cancer Study (AOCS) Group, be entrapment of epithelial cells within proliferating stroma collaborators, and hospitals involved in AOCS can be found at http://www. aocstudy.org. that subsequently drives the formation of cystic masses. Under these circumstances we propose that growth factors Grant Support secreted by the fibroma, or inhibitory factors no longer secreted by the fibroma, promote proliferation of the epi- This work was supported by a grant (ID 628630) from the National thelium, which subsequently increases the likelihood of the Health and Medical Research Council of Australia (NHMRC). The AOCS was epithelium acquiring or selecting for genetic alterations supported by the U.S. Army Medical Research and Materiel Command under DAMD17-01-1-0729, The Cancer Council Tasmania, The Cancer Founda- leading to neoplasia (Fig. 3). The capacity for genetically tion of Western Australia, and the National Health and Medical Research altered fibroblasts to induce neoplasia in epithelial cells has Council of Australia, (NHMRC; IDs 40028 and 400413). been shown in a number of mouse models (49, 50). This The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked alternative pathway may exist independently of the classic advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate pathways of low-grade serous tumors characterized by the this fact. presence of activating mutations in the oncogenes KRAS BRAF and in the epithelium and may even explain some Received August 12, 2011; revised September 28, 2011; accepted high-grade serous and endometrioid tumors. September 29, 2011; published OnlineFirst October 5, 2011.

References 1. Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancer 10. Kurman RJ, Hedrick Ellenson L, Ronnett BM, editors. Blaustein's pathol- statistics. CA Cancer J Clin 2011;61:69–90. ogy of the female genital tract. New York: Springer; 2011. p. 1173. 2. Averette HE, Janicek MF, Menck HR. The National Cancer Data Base 11. Kobel M, Kalloger SE, Boyd N, McKinney S, Mehl E, Palmer C, et al. report on ovarian cancer. American College of Surgeons Commission on Ovarian carcinoma subtypes are different diseases: implications for Cancer and the American Cancer Society. Cancer 1995;76:1096–103. biomarker studies. PLoS Med 2008;5:e232. 3. Jemal A, Murray T, Ward E, Samuels A, Tiwari RC, Ghafoor A, et al. 12. Gilks CB, Ionescu DN, Kalloger SE, Kobel M, Irving J, Clarke B, et al. Cancer statistics, 2005. CA Cancer J Clin 2005;55:10–30. Tumor cell type can be reproducibly diagnosed and is of independent 4. Takayama T, Katsuki S, Takahashi Y, Ohi M, Nojiri S, Sakamaki S, et al. prognostic signiﬁcance in patients with maximally debulked ovarian Aberrant crypt foci of the colon as precursors of adenoma and cancer. carcinoma. Hum Pathol 2008;39:1239–51. N Engl J Med 1998;339:1277–84. 13. Shih Ie M, Kurman RJ. Ovarian tumorigenesis: a proposed model 5. Blaustein A, Kaganowicz A, Wells J. Tumor markers in inclusion cysts based on morphological and molecular genetic analysis. Am J Pathol of the ovary. Cancer 1982;49:722–6. 2004;164:1511–8. 6. Singer G, Kurman RJ, Chang HW, Cho SK, Shih Ie M. Diverse 14. Ho CL, Kurman RJ, Dehari R, Wang TL, Shih Ie M. Mutations of BRAF tumorigenic pathways in ovarian serous carcinoma. Am J Pathol and KRAS precede the development of ovarian serous borderline 2002;160:1223–8. tumors. Cancer Res 2004;64:6915–8. 7. Pothuri B, Leitao MM, Levine DA, Viale A, Olshen AB, Arroyo C, et al. 15. Teneriello MG, Ebina M, Linnoila RI, Henry M, Nash JD, Park RC, et al. Genetic analysis of the early natural history of epithelial ovarian p53 and Ki-ras gene mutations in epithelial ovarian neoplasms. Cancer carcinoma. PLoS One 2010;5:e10358. Res 1993;53:3103–8. 8. Korner M, Burckhardt E, Mazzucchelli L. Different proportions of 16. Singer G, Oldt R III, Cohen Y, Wang BG, Sidransky D, Kurman RJ, et al. aneusomic cells in ovarian inclusion cysts associated with serous Mutations in BRAF and KRAS characterize the development of low- borderline tumours and serous high-grade carcinomas support dif- grade ovarian serous carcinoma. J Natl Cancer Inst 2003;95:484–6. ferent pathogenetic pathways. J Pathol 2005;207:20–6. 17. Sieben NL, Macropoulos P, Roemen GM, Kolkman-Uljee SM, Jan 9. Barakat RR, Markman M, Randall ME, editors. Principles and practice Fleuren G, Houmadi R, et al. In ovarian neoplasms, BRAF, but not of gynecologic oncology. Philadelphia: Lippincott Williams & Wilkins; KRAS, mutations are restricted to low-grade serous tumors. J Pathol 2009. 2004;202:336–40.

www.aacrjournals.org Clin Cancer Res; 17(23) December 1, 2011 7281

Hunter et al.

18. Kuo KT, Guan B, Feng Y, Mao TL, Chen X, Jinawath N, et al. Analysis of ysis of 307 cases with histogenetic implications. Virchows Arch DNA copy number alterations in ovarian serous tumors identifies new 2009;454:677–83. molecular genetic changes in low-grade and high-grade carcinomas. 35. Mayr D, Hirschmann A, Lohrs U, Diebold J. KRAS and BRAF mutations Cancer Res 2009;69:4036–42. in ovarian tumors: a comprehensive study of invasive carcinomas, 19. Singer G, Stohr R, Cope L, Dehari R, Hartmann A, Cao DF, et al. borderline tumors and extraovarian implants. Gynecol Oncol Patterns of p53 mutations separate ovarian serous borderline tumors 2006;103:883–7. and low- and high-grade carcinomas and provide support for a new 36. Seidman JD, Mehrotra A. Benign ovarian serous tumors: a re-evalu- model of ovarian carcinogenesis: a mutational analysis with immuno- ation and proposed reclassification of serous "cystadenomas" and histochemical correlation. Am J Surg Pathol 2005;29:218–24. "cystadenofibromas". Gynecol Oncol 2005;96:395–401. 20. Crum CP, Drapkin R, Miron A, Ince TA, Muto M, Kindelberger DW, et al. 37. Fletcher JA, Gibas Z, Donovan K, Perez-Atayde A, Genest D, Morton The distal fallopian tube: a new model for pelvic serous carcinogenesis. CC, et al. Ovarian granulosa-stromal cell tumors are characterized by Curr Opin Obstet Gynecol 2007;19:3–9. trisomy 12. Am J Pathol 1991;138:515–20. 21. Tone AA, Begley H, Sharma M, Murphy J, Rosen B, Brown TJ, et al. 38. Leung WY, Schwartz PE, Ng HT, Yang-Feng TL. Trisomy 12 in benign Gene expression profiles of luteal phase fallopian tube epithelium from fibroma and granulosa cell tumor of the ovary. Gynecol Oncol BRCA mutation carriers resemble high-grade serous carcinoma. Clin 1990;38:28–31. Cancer Res 2008;14:4067–78. 39. Mrozek K, Nedoszytko B, Babinska M, Mrozek E, Hrabowska M, 22. Carlson J, Roh MH, Chang MC, Crum CP. Recent advances in the Emerich J, et al. Trisomy of chromosome 12 in a case of thecoma of understanding of the pathogenesis of serous carcinoma: the concept the ovary. Gynecol Oncol 1990;36:413–6. of low- and high-grade disease and the role of the fallopian tube. Diagn 40. Pejovic T, Heim S, Mandahl N, Elmfors B, Floderus UM, Furgyik S, et al. Histopathol (Oxf) 2008;14:352–65. Trisomy 12 is a consistent chromosomal aberration in benign ovarian 23. Kindelberger DW, Lee Y, Miron A, Hirsch MS, Feltmate C, Medeiros F, tumors. Genes Chromosomes Cancer 1990;2:48–52. et al. Intraepithelial carcinoma of the fimbria and pelvic serous carci- 41. Persons DL, Hartmann LC, Herath JF, Keeney GL, Jenkins RB. Fluo- noma: Evidence for a causal relationship. Am J Surg Pathol rescence in situ hybridization analysis of trisomy 12 in ovarian tumors. 2007;31:161–9. Am J Clin Pathol 1994;102:775–9. 24. Bryan EJ, Watson RH, Davis M, Hitchcock A, Foulkes WD, Campbell 42. Yang-Feng TL, Li SB, Leung WY, Carcangiu ML, Schwartz PE. Trisomy IG. Localization of an ovarian cancer tumor suppressor gene to a 0.5- 12 and K-ras-2 amplification in human ovarian tumors. Int J Cancer cM region between D22S284 and CYP2D, on chromosome 22q. 1991;48:678–81. Cancer Res 1996;56:719–21. 43. Liang JC, Kurzrock R, Gutterman JU, Gallick GE. Trisomy 12 correlates 25. Merritt MA, Green AC, Nagle CM, Webb PM. Talcum powder, chronic with elevated expression of p21 ras in a human adenosquamous pelvic inflammation and NSAIDs in relation to risk of epithelial ovarian carcinoma of the lung. Cancer Genet Cytogenet 1986;23:183–8. cancer. Int J Cancer 2008;122:170–6. 44. Pickhardt PJ, Hanson ME. Incidental adnexal masses detected at low- 26. Wang Y, Carlton VE, Karlin-Neumann G, Sapolsky R, Zhang L, Moor- dose unenhanced CT in asymptomatic women age 50 and older: head M, et al. High quality copy number and genotype data from FFPE implications for clinical management and ovarian cancer screening. samples using Molecular Inversion Probe (MIP) microarrays. BMC Med Radiology 2010;257:144–50. Genomics 2009;2:8. 45. Bhowmick NA, Neilson EG, Moses HL. Stromal fibroblasts in cancer 27. Hardenbol P, Yu F, Belmont J, Mackenzie J, Bruckner C, Brundage T, initiation and progression. Nature 2004;432:332–7. et al. Highly multiplexed molecular inversion probe genotyping: over 46. Bhowmick NA, Chytil A, Plieth D, Gorska AE, Dumont N, Shappell S, 10,000 targeted SNPs genotyped in a single tube assay. Genome Res et al. TGF-beta signaling in fibroblasts modulates the oncogenic 2005;15:269–75. potential of adjacent epithelia. Science 2004;303:848–51. 28. Wiegand KC, Shah SP, Al-Agha OM, Zhao Y, Tse K, Zeng T, et al. 47. Ohuchida K, Mizumoto K, Murakami M, Qian LW, Sato N, Nagai E, et al. ARID1A mutations in endometriosis-associated ovarian carcinomas. Radiation to stromal fibroblasts increases invasiveness of pancreatic N Engl J Med 2010;363:1532–43. cancer cells through tumor-stromal interactions. Cancer Res 2004; 29. Qiu W, Hu M, Sridhar A, Opeskin K, Fox S, Shipitsin M, et al. No 64:3215–22. evidence of clonal somatic genetic alterations in cancer-associated 48. Barcellos-Hoff MH, Ravani SA. Irradiated mammary gland stroma fibroblasts from human breast and ovarian carcinomas. Nat Genet promotes the expression of tumorigenic potential by unirradiated 2008;40:650–5. epithelial cells. Cancer Res 2000;60:1254–60. 30. Roy WJ Jr, Watson RH, Hitchcock A, Campbell IG. Frequent loss of 49. Kuperwasser C, Chavarria T, Wu M, Magrane G, Gray JW, Carey L, heterozygosity on chromosomes 7 and 9 in benign epithelial ovarian et al. Reconstruction of functionally normal and malignant human tumors. Oncogene 1997;15:2031–5. breast tissues in mice. Proc Natl Acad Sci U S A 2004;101: 31. Thomas NA, Neville PJ, Baxter SW, Campbell IG. Genetic analysis of 4966–71. benign ovarian tumors. Int J Cancer 2003;105:499–505. 50. Bhowmick NA, Chytil A, Plieth D, Gorska AE, Dumont N, Shappell S, 32. Antill YC, Mitchell G, Johnson SA, Devereux L, Milner A, Phillips KA, et al. TGF-beta signaling in fibroblasts modulates the oncogenic et al. Loss of heterozygosity analysis in ductal lavage samples from potential of adjacent epithelia. Science 2004;303:848–51. BRCA1 and BRCA2 carriers: a cautionary tale. Cancer Epidemiol 51. HelouK,Padilla-NashH,WangsaD,KarlssonE,OsterbergL, Biomarkers Prev 2006;15:1396–8. Karlsson P, et al. Comparative genome hybridization reveals spe- 33. Cheng EJ, Kurman RJ, Wang M, Oldt R, Wang BG, Berman DM, et al. cific genomic imbalances during the genesis from benign through Molecular genetic analysis of ovarian serous cystadenomas. Lab borderline to malignant ovarian tumors. Cancer Genet Cytogenet Invest 2004;84:778–84. 2006;170:1–8. 34. Pradhan M, Davidson B, Trope CG, Danielsen HE, Abeler VM, Risberg 52. Micci F, Haugom L, Ahlquist T, Andersen HK, Abeler VM, Davidson B, B. Gross genomic alterations differ between serous borderline tumors et al. Genomic aberrations in borderline ovarian tumors. J Transl Med and serous adenocarcinomas–an image cytometric DNA ploidy anal- 2010;8:21.

7282 Clin Cancer Res; 17(23) December 1, 2011 Clinical Cancer Research

Copy Number Aberrations in Benign Serous Ovarian Tumors: A Case for Reclassification?

Sally M. Hunter, Michael S. Anglesio, Raghwa Sharma, et al.

Clin Cancer Res 2011;17:7273-7282. Published OnlineFirst October 5, 2011.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-11-2080

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2011/10/04/1078-0432.CCR-11-2080.DC1

Cited articles This article cites 50 articles, 12 of which you can access for free at: http://clincancerres.aacrjournals.org/content/17/23/7273.full#ref-list-1

Citing articles This article has been cited by 2 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/17/23/7273.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 Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/17/23/7273. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.