Allele loss from large regions of chromosome 17 is common only in certain histological subtypes of ovarian carcinomas.

Using a panel of ten polymorphic markers, we examined the frequency of loss of heterozygosity (LOH) on chromosome 17 in 55 sporadic ovarian tumours. LOH on 17p and 17q was observed to be 50% and 62% respectively. LOH at D17S5 was detected in 24/36 (67%) of malignant cases and in 19/43 (44%) at TP53; the marker D17S855 intragenic to the BRCA1 gene showed allele loss in 50% (20/40) cases. The data presented here suggest that loss of the whole chromosome 17 is a relatively frequent event (30%) in ovarian carcinomas and this observation is especially frequent for serous, transitional cell and anaplastic histological subtypes. Mucinous and endometrioid ovarian tumours showed only short interstitial deletions (4/11, 36%). The overall frequency of the short deletions was relatively low (7/43, 16%) in our panel of carcinomas. Amplification of c-erbB-2/neu oncogene was detected in 32% (11/34) of the carcinomas tested; the gene was amplified only in those histological subtypes in which high incidence of LOH on chromosome 17 was observed, and was associated with advanced stages of the disease. We conclude that different histological types of tumour may have different aetiological mechanisms, and tumour-suppressor genes on chromosome 17 might be associated specifically with serous and transitional cell ovarian carcinomas.

Molelcular genetic analysis of ovarian carcinomas has revealed a significant role for chromosome 17 in pathogenesis of ovarian malignancies. These studies have shown that loss of heterozygosity (LOH) for regions of chromosome 17 is a frequent event, probably indicating the inactivation of suppressor genes present on this chromosome (Eccles et al., 1990;Lee et al., 1990;Russell et al., 1990;Foulkes et al., 1991;Phillips et al., 1993). The search for loss of constitutional heterozygosity with polymorphic genetic markers is now a widely accepted approach to indicate areas on the genome where inactivation of tumour-suppressor genes may occur.
In addition to the studies addressing the importance of specific tumour-suppressor genes, several groups examined the frequency of oncogenes activated in ovarian carcinomas. Amplification of the c-erbB-2/neu oncogene was observed most frequently, and was associated with advanced stages and poor clinical outcome (Slamon et al., 1989). In order to contribute to the clarification of the biological significance of genetic alterations on chromosome 17 in ovarian cancer, in this study we further examined the amplification of c-erbB-2 oncogene and the frequencies of losses of heterozygosity at ten different loci on chromosome 17 in 50 epithelial and in five non-epithelial ovarian tumours.

Samples
Fresh tumour tissue samples were collected from consenting patients undergoing surgery for ovarian cancer, who had received no prior therapy. Samples were collected in dry ice and stored at -80°C until processed. Histopathological classification of the ovarian tumours was carried out according to the WHO classification. The distribution of the epithelial tumours was as follows: 24 serous, six mucinous, seven endometrioid, one clear cell, six transitional cell, three anaplastic and three mixed cell tumours. Stages of the disease were assigned according to the classification scheme accepted by the general assembly of the International Federation of Gynecologists and Obstetricians (FIGO). DNA extraction, Southern analysis DNA was extracted by standard methods from fresh-frozen samples and from peripheral lymphocytes from all patients. Southern analysis was performed by standard techniques. For typing marker D17S4, RsaI-digested genomic DNA was hybridised with the whole linearised plasmid (pTHH59). For detection of c-erbB-2 oncogene amplification, PstI-digested ovarian carcinoma DNA was hybridised with labelled human c-erbB-2 probe, washed, autoradiographed, and reprobed with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) probe, specific for a single-copy gene.

Microsatellite markers
Eight of the ten polymorphic markers (TP53, D17S261, D17S250, THRAI, D17S855, D17S579, D17S588 and NME1) detected (CA)n dinucleotide repeat polymorphisms, and two of them were VNTR markers (Dl7S5 and D17S4). The samples were scored for LOH by comparing the autoradiographic signals of the corresponding blood and tumour tissue samples. 50 mM potassium chloride, 10 mM Tris, pH 8.3, and 1.5 U AmpliTaq DNA Polymerase (Perkin Elmer Cetus). The 5' primers of microsatellite markers were end labelled with [y-32P]ATP using T4 polynucleotide kinase B. The samples Allele loss from chromosome 17 in ovarian cancer J Papp et al $ 1593 were amplified in 28-35 cycles, each containing a denaturation step (1 min at 95°C), an annealing step (1 or 2 min at the appropriate annealing temperature) and an extension step (1 min at 72°C). For typing the TP53 marker we used a two-    step PCR amplification protocol decribed by Jones and Nakamura (1992). In all cases, PCR cycles were preceded by an initial denaturation step (10 min at 95°C) and followed by an elongation step (7 min at 72°C). PCR products of microsatellite markers were run on a standard sequencing gel using an M13 sequencing reaction as size marker. After fixation and drying, gels were autoradiographed for 1 -3 days at room temperature. Amplified fragments of the YNZ22 marker with VNTR polymorphism were run on 2% agarose (SeaKem) gel and visualised by ethidium bromide staining.
Statistical analysis All comparisons between groups and/or parameters were performed using Fisher's exact t-test. One-tailed P-values<0.05 were considered statistically significant.

Results
After characterising all 55 cases of primary tumours of the ovary by histology and grade, DNA extracted from tumours and corresponding normal DNAs were screened for allele loss of both arms of chromosome 17 using ten polymorphic markers. The informativity of the markers and the frequencies of LOHs are summarised in Table I. The pattern of losses is presented in Figure 1. In non-epithelial ovarian tumours allelic deletion was relatively common; three out of five non-epithelial malignant tumours showed deletions affecting only one marker (TP53) in one case and long chromosomal fragments in the two other cases. From our small panel of informative benign and borderline epithelial tumours, only the tumour with borderline malignancy showed LOH, while no allele loss was seen in any of the benign tumours.
Out of 44 carcinoma samples, 75% presented allele loss from chromosome 17. These results indicate strongly significant correlation (P= 0.002) between malignancy and LOH.
Allelic deletion of p53 was observed in 50% (18/39) of malignant epithelial tumours. The frequent occurrence (23/32, 72%) of LOH at D17S5 may imply the presence of a tumoursuppressor gene at the region telomeric to p53. The allele losses involved the BRCAJ gene defined by the intragenic marker D17S855 in 51% (22/43) of the malignant cases tested. Table II shows loss of heterozygosity according to the histopathological subtypes. There was a statistically significant difference (P=0.006) in frequencies of chromosome 17 LOH between serous and transitional cell vs mucinous and endometrioid histological groups of malignant ovarian tumours. The overwhelming majority of losses was seen in the subtypes of serous (16/20, 80%), transitional cell (6/6, 100%), anaplastic (3/3, 100%) and mixed cell (3/3, 100%) ovarian carcinomas. In these subtypes not only the incidence of LOH was higher, but also longer chromosomal regions were involved. At least 19 cases out of 43 carcinomas (44%) were suggestive of the loss of the whole long arm, whereas 37% (16/43) showed loss of the short arm. (As an example, pattern of allelic losses of patient no. 57 is shown in Figure  2). Loss of the entire chromosome 17 was found in 30% (13/ 44) of carcinomas examined. In contrast, only one of four mucinous and three of six endometrioid tumours presented allelic deletions, affecting only one or two markers in each case. The overall frequency of short, interstitial deletions was relatively low (7/43, 16%) in our panel of malignant tumours.
To elucidate the genetic imbalance of ovarian carcinoma cells further, amplification of the c-erbB-2 oncogene, which is associated with poor prognosis in ovarian as well as in breast carcinomas, was also evaluated (Figures 1 and 3). The frequency of amplification of this gene in our ovarian tumours was determined and related to the clinical stage and histological grade of the disease. We found 2-5-fold amplification of the c-erbB-2 oncogene in 11 of 34 (32%) of the carcinomas and neither benign tumours nor non-epithelial malignant cases showed c-erbB-2 amplification (Table II). All tumours, except one containing amplified c-erbB-2, were of advanced stage, and amplification of c-erbB-2 was associated (P= 0.011) with higher grade. c-erbB-2 was found to be amplified only in those histological subtypes, in which high incidence of LOH on chromosome 17 was observed.  (25) a LOH affecting at least one marker on chromosome 17. whole copy of the chromosome 17 suggests the possible involvement of multiple chromosome 17 loci in the pathogenesis in some ovarian tumours. This is consistent with the presence of tumour-suppressor genes on both arms of this chromosome, including p53, BRCAJ and a potential tumour-suppressor gene distal to BRCAI (Jacobs et al., 1993;Godwin et al., 1994).
Specifically, all losses of the whole chromosome 17 and the vast majority of losses of large chromosomal regions were detected in serous, anaplastic, transitional cell or mixed cell ovarian carcinomas. It is noteworthy that serous carcinoma samples without any loss ( 52 and 60 demonstrate more than 2-fold amplification. The same LOH on the short arm of the chromosome, including the filter was reprobed with the GAPDH probe (lower panel); Co, p53 locus, was observed in ovarian carcinomas by several normal lymphocyte control.
groups (Coles et al., 1990;Eccles et al., 1992b;Kohler et al., 1993). In our studies there was 50% LOH at p53. Allelic losses and mutations of the p53 gene are common genetic Discussion events in ovarian cancer (Okamoto et al., 1991;Millner et al., 1993), indicating a direct involvement of p53 in ovarian Very frequent LOH occurred at all chromosome 17 loci malignancies. On the basis of LOH studies, several reports examined in this study. In contrast to reports on chromosome have suggested that in addition to p53 there may be a gene 17 in breast cancer (Lindblom et al., 1993) and other telomeric of p53 (at 17pl3.3), which is acting as a tumour chromosomes involved in ovarian carcinogenesis (e.g. suppressor or a regulator of p53 expression in breast an-d chromosome 6) (Wan et al., 1994), in which LOH affects ovarian carcinogenesis (Tsao et al., 1991;Wales et al., 1995; only specific regions, in a fairly high proportion of our Stack et al., 1995). Coles et al. (1990) reported a significantly carcinomas the loss appeared to involve the whole higher frequency of LOH at 17pl3.3 than at the p53 locus. chromosome (30%). These results are in line with the Although D17S5 and p53 deletions were observed together in c-erbB-2 GAPDH 56% of the cases in our studies, eight tumours showed both loss of D17S5 and retention of p53, which may indicate a second putative tumour-suppressor gene locus in this region. (A typical case is shown in Figure 2.) Allelic losses on 17p have been observed in various malignancies, but LOH on 17q appears to be more specific for breast and ovarian carcinomas. The familial breast/ ovarian cancer locus (BRCA1) is mapped to 17q21, and the BRCAI gene has already been isolated (Miki et al., 1994). In a multicentre study, we recently reported LOH in 86% of familial breast and ovarian tumours, which invariably involved the wild-type allele (Cornelis et al., 1995). These results strongly support the case that BRCAJ is a tumoursuppressor gene and the loss of heterozygosity is greatly favoured to inactivate it fully. Somatic point mutations in BRCAJ are relatively infrequent in sporadic breast and ovarian cancer, but deletion of one BRCAJ allele occurs in approximately 50% of sporadic breast and 70% of sporadic ovarian cancer (Futreal et al., 1994;Merajver et al., 1995;Holt et al., 1996). The long arm of chromosome 17 displayed LOH in 50% of our malignant cases in the BRCAI gene, and 63% of the cases with the marker close to the NMEl metastasis-suppressor gene. Most of the cases of LOH on 17q included the region in which the BRCAJ gene is located. However, three cases with telomeric losses showed no deletion in this region. In our samples, the frequency of allele loss was always higher when telomeric markers were used. However, the LOH detected with markers close to the telomeric regions (like D17S5 and D17S4) may be indicative of chromosome 17 telomeric losses without any association with specific genes. Telomeric deletion on 17q, which has been shown to be associated with chromosomal instabilities in a number of human tumours, including ovarian cancer (Hastie et al., 1990), can also contribute to loss of large chromosome fragments. Saito et al. (1992,1993) found a significant difference in the frequency of LOH at 6q and 17q21.3 among three different histopathological groups: tumours of the serous type showed LOH more often than did mucinous or clear cell types. Our results reported here demonstrate that losses affecting the whole chromosome 17 or long fragments of either arm are characteristic only for serous, transitional cell and anaplastic groups, while mucinous, endometrioid and clear cell carcinomas either show no LOH or only interstitial allelic losses. Most recent support for preferential involvement of serous ovarian carcinomas in chromosome 17 loss came from Pieretti et al. (1995).
In our studies the c-erbB-2/neu locus was found to be amplified in 32% of the malignant cases and c-erbB-2 amplification was more characteristic of higher histological grade and clinical stage. These results are in agreement with the report of Slamon et al. (1989), who first demonstrated a relationship between c-erbB-2 amplification and poor prognosis for ovarian as well as breast carcinomas. The c-erbB-2 gene was amplified only in those histological subtypes (serous, anaplastic, transitional cell and mixed cell) in which high incidence of LOH occurred. In 10 out of 11 cases, both amplification and loss of the same region (D17S250, THRAI) were observed. The term allelic imbalance, introduced by Devilee and Cornelisse (1994), would provide a better description of such DNA changes as it permits interpretation of imbalance of allelic signals, irrespective of whether it is an allelic loss or gain.
The data presented here further suggest that allelic imbalance of chromosome 17 is an important genetic event in epithelial ovarian carcinogenesis. Our results also indicate that there might be different genetic pathways in the aetiology of histological subtypes of ovarian cancer.