Oncogene (1998) 16, 197 ± 202 1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 A one centimorgan deletion unit on chromosome Xq12 is commonly lost in borderline and invasive epithelial ovarian tumors MI Edelson1, CC Lau4, CV Colitti1, WR Welch2, DA Bell3, RS Berkowitz1 and SC Mok1 1Laboratory of Gynecologic Oncology, Division of Gynecologic Oncology, Department of Obstetrics, Gynecology and Reproductive Biology, and 2Department of Pathology, Brigham and Women's Hospital, Boston, Massachusetts, 02115; 3Department of Pathology, Massachusetts General Hospital, Boston, Massachusetts 02114; 4Division of Hematology-Oncology, Texas Children's Hospital, Houston, Texas, 77030, USA We have used polymerase chain reaction (PCR) have been unable to detect any signi®cant level of ampli®cation of tandem repeats to study the pattern of genetic alterations speci®c to borderline tumors allelic loss on chromosome X11.2-q12 in borderline and emphasizing the possible separate pathogenesis of invasive epithelial ovarian tumors. Using eight micro- these tumors (Rodabaugh et al., 1995b; Wertheim et satellite markers spanning Xq11.2-q12, 41 borderline and al., 1994; Tangir et al., 1996; Rodabaugh et al., 1995a; 65 invasive ovarian tumors, together with their corre- Dodson et al., 1993; Kim et al., 1994). Recently, Zheng sponding normal tissues, were analysed. The highest et al. have demonstrated that 50% of informative percentage of loss of heterozygosity (LOH) was observed borderline tumors (eight of 16 informative cases) and at the DXS1194 locus in borderline tumors (four of 16 approximately 44% of high grade IEOC displayed informative cases, 25%) and at the androgen receptor LOH at the androgen receptor (AR) locus in the (AR) locus in invasive epithelial ovarian tumors (18 of 47 proximal portion of chromosome Xq (Cheng et al., informative cases, 38%). X chromosome activation 1996). These results suggest that LOH in this region studies performed in cases with LOH at the AR locus may be important for the development of borderline showed that the allelic loss at the AR locus is not tumors and high grade invasive epithelial ovarian con®ned to the inactive allele. A one centimorgan region carcinomas. including the AR locus and ¯anked by the primers In this study, we report the construction of a DXS1161 and PGK1P1 was identi®ed as the smallest detailed deletion map of chromosome Xq11.2-q12 common loss region in both borderline and invasive from 41 borderline tumors and 65 invasive epithelial epithelial ovarian tumors. ovarian carcinomas using eight microsatellite markers in an eort to more accurately localize a region(s) Keywords: ovary; cancer; borderline; loss of hetero- which may be involved in the development of zygosity; chromosome Xq borderline ovarian tumors and invasive EOC. Results Introduction LOH in Xq11.2-Xq12 Borderline ovarian tumors, also known as ovarian epithelial tumors of low malignant potential, are a A set of eight microsatellite markers spanning the distinct entity within the classi®cation of ovarian region from Xq11.2 to Xq12 (Figure 1) was used to neoplasms. These tumors have a histological appear- perform the LOH study on 41 borderline ovarian ance and biologic behavior which are intermediate tumors and 65 invasive epithelial ovarian carcinomas between benign and frankly malignant ovarian with various stages and grades. All cases showed neoplasms (Scully, 1982). While these tumors are interstitial LOH for several of the microsatellite loci cytologically malignant with evidence of cellular tested. LOH aecting at least one locus was observed pleomorphism, they do not exhibit any stromal in 29% of the borderline tumors (12 of 41 informative invasion. A prolonged survival is expected with these cases) and 34% of the invasive epithelial carcinomas tumors even without the use of chemotherapy (22 of 65 informative cases). The LOH results derived (Bostwick et al., 1986; Barakat, 1994; Leake, 1992; from paran embedded tissue and frozen materials Leake et al., 1992; Casey et al., 1993). were not dierent statistically, and thus were combined Genetic studies have been used to understand the in the analysis of both borderline ovarian tumors and biology of ovarian cancers. Loss of heterozygosity invasive epithelial ovarian carcinomas. Figure 2 shows (LOH) studies have been used to detect allelic losses of autoradiographs of ®ve borderline tumor cases speci®c chromosomes in order to identify possible depicting LOH at the AR locus in the tumor lane tumor suppressor genes which may be relevant to the with the normal control in the adjacent lane. development of ovarian cancer. While many of these studies have focused on invasive epithelial ovarian LOH in borderline tumors carcinomas, only a few studies have examined genetic alterations of borderline ovarian tumors. These studies The percentage of tumors with allelic loss at each locus for borderline tumors is shown in Figure 1. The highest Correspondence: SC Mok percentage of LOH was observed at DXS1213 (four of Received 8 February 1997; revised 5 August 1997; accepted 6 August 16 informative cases, 25%) and at the AR locus (seven 1997 of 30 informative cases, 23%). All other loci showed a Deletion mapping of chromosome Xq in ovarian cancer MI Edelson et al 198 Figure 1 Summary of the loci, map position, and frequency of loss of heterozygosity at Xq11.2-q12 in borderline ovarian tumors and invasive epithelial ovarian carcinomas mately one centimorgan, is located between the markers DXS1161 and PGK1P1. The identi®cation of B5 B43 333 466 470 this region was based on the study of cases 315, 317, T N T N T N T N T N 416 and 490 (Figure 4). AR LOH with respect to grade, stage and histological type Clinicopathological characteristics of the tumors were analysed based on allelic loss at the AR locus (Table Figure 2 Autoradiographs of PCR ampli®cations of DNA from 1). No signi®cant dierence was observed between paran blocks (patients B5 and B43) and frozen materials LOH rate at the androgen receptor locus and stage or (patients 333, 466, 470) for borderline ovarian tumors at the androgen receptor (AR) locus. T, tumor DNA; N, normal DNA; tumor type for borderline tumors. Similar analysis for Allelic loss is visualized in the tumor lanes invasive epithelial ovarian carcinomas did reveal a statistically signi®cant dierence of allelic loss at the AR locus for Stage III/IV tumors (P=0.04) and serous much lower percentage of LOH. A one centimorgan tumors (P=0.02). Comparison of LOH rate at the AR region including the AR locus and ¯anked by the locus between borderline tumors (23%), low grade primers DXS1161 and PGK1P1 was identi®ed as the carcinomas (32%), and high grade carcinomas (53%) smallest common loss region using selected cases which did not reveal any statistical dierence (P=0.10), showed restricted LOH patterns (Figure 3). Cases B5, though a trend towards a greater frequency of LOH 466, and 543 speci®cally show LOH at the AR locus at the AR locus with less dierentiated tumors was and no LOH at the ¯anking primers DXS1161 and observed. PGK1P1. No selective loss of the inactive allele at the AR locus LOH in invasive epithelial ovarian carcinomas Selected paired normal and tumor DNA from cases Similar to the borderline tumors, the AR locus showed which showed LOH at the AR locus were subjected to the highest percentage of allelic loss (18 of 47 HpaII restriction enzyme digestion to evaluate whether informative cases, 38%, Figure 1). However, the the lost allele is active or inactive. After digestion, the ¯anking primers showed a higher percentage of LOH remaining allele of cases 324, 336, 490, 528, 545 and in the invasive epithelial ovarian carcinomas as 558 disappeared after digestion suggesting that in these compared to the borderline tumors. Five cases (ES-3, cases the inactive AR allele was lost (Figure 5a). In 351, 384, 528 and 545) showed allelic loss at almost all contrast, the remaining allele of cases 332, 384, and 386 the loci suggesting that these samples had LOH in the could still be observed, suggesting that the active AR whole Xq11.2-q12 region. Therefore, these cases were allele was lost in these cases (Figure 5b). When DNA not informative for the localization of the smallest from normal tissue from cases 324, 332, 386, 490, 528, region of allelic loss. The remaining 17 cases showed and 558 were digested with HpaII, a skewing pattern of restricted LOH patterns which allowed us to de®ne the the two AR alleles was observed. This was particularly smallest common loss region. This region, approxi- evident in case 386 suggesting a relatively selective X- Deletion mapping of chromosome Xq in ovarian cancer MI Edelson et al 199 Figure 3 Deletion map of 12 borderline ovarian tumors. Shaded region indicates loss of region. Primers used in the PCR reaction are listed above. For map of location of the primers, see Figure 1. L, loss of heterozygosity. N, no loss. U, uninformative Figure 4 Deletion map of 22 invasive epithelial ovarian carcinomas. Shaded region indicates loss of region. Primers used in the PCR reaction are listed above. For map of location of the primers, see Figure 1. L, loss of heterozygosity. N, no loss. U, uninformative chromosome inactivation in the normal tissue of these Table 1 Relationship between allelic loss for androgen receptor loci cases. and clinicopathologic characteristics in 30 borderline ovarian tumors and 47 invasive epithelial ovarian carcinomas Invasive The AR allele was not totally deleted in a majority of epithelial ovarian cases that showed LOH Clinico- Borderline tumors carcinomas pathologic LOH No loss LOH No loss To evaluate whether loss of heterozygosity observed at characteristics (n=7) (n=23) P (n=18) (n=29) P the AR locus was due to total allelic deletion or Stage deletion with the replacement by the remaining allele, I/II 7 18 1 13 we performed multiplex PCR.