Amplification Of

Ampliﬁcation of HSD17B1 and ERBB2 in primary breast cancer

Cecilia Gunnarsson1, Marie Ahnstro¨ m1, Kristina Kirschner1, Birgit Olsson1, Bo Nordenskjo¨ ld1, Lars Erik Rutqvist2, Lambert Skoog3 and Olle Sta˚ l*,1

1Department of Biomedicine and Surgery, Division of Oncology, Faculty of Health Sciences, Linko¨ping University, SE-581 85 Linko¨ping, Sweden; 2Department of Oncology, Huddinge University Hospital, SE-14 186 Stockholm, Sweden; 3Division of Cytology, Karolinska Hospital, SE-10401 Stockholm, Sweden

Estrogens play a crucial role in the development of breast androgens to estrogens, and the other utilizing sulfatase, cancer. Estradiol can be produced in the breast tissue in which converts estrone sulfate into estrone. Previous situ, and one of the enzymes involved in this process is studies have suggested that estradiol can be produced in 17b-hydroxysteriod dehydrogenase (17b-HSD) type 1 the same organ where it exerts its biological response. that catalyzes the interconversion of estrone (E1) to the This is also in agreement with the fact that breast cancer biologically more potent estradiol (E2). The gene coding tissue possesses all the enzymes necessary for the for 17b-HSD type 1 (HSD17B1) is located at 17q12–21, bioformation of estradiol (Yue et al., 1998; Purohit close to the more studied ERBB2 and BRCA1. The aim of et al., 2002). Among these enzymes are 17b-hydroxys- this study was to investigate if HSD17B1 shows an altered teriod dehydrogenases (17b-HSD), which control the gene copy number in breast cancer. We used real-time final step in the formation of androgens and estrogens PCR and examined 221 postmenopausal breast tumors for (Labrie et al., 1997). The 17b-HSD type 1 enzyme uses amplification of HSD17B1 and ERBB2. In all, 32 tumors NADPH as a cofactor and catalyzes the interconversion (14.5%) showed amplification of HSD17B1 and 21% of the weak estrogen, estrone (E1), to the biologically were amplified for ERBB2. Amplification of the two genes more potent estradiol (E2). 17b-HSD type 2 uses was correlated (P ¼ 0.00078) and in 14 tumors (44%) with NAD+ as a cofactor and catalyzes the oxidation of amplification of HSD17B1, ERBB2 was co amplified. The testosterone and estradiol to form androstendione and patients with amplification in at least one of the genes had estrone, respectively (Miettinen et al., 1996; Vihko et al., a significantly worse outcome than patients without 2001). (P ¼ 0.0059). For estrogen receptor (ER)-positive patients A recent study showed that the ratio of bioactive who received adjuvant tamoxifen, amplification of estradiol to estrone (E2/E1) in breast tumors as well as HSD17B1 was related to decreased breast cancer survival the expression level of 17b-HSD type 1 mRNA was (P ¼ 0.017), whereas amplification of ERRB2 was not. higher among postmenopausal than premenopausal Amplification of HSD17B1 might be an indicator of patients (Miyoshi et al., 2001). A few immunohisto- adverse prognosis among ER-positive patients, and chemical studies of 17b-HSD type 1 in human breast possibly a mechanism for decreased benefit from tamox- cancer have been reported, and these suggest that 17b- ifen treatment. HSD 1 may play an important role in the in situ Oncogene (2003) 22, 34–40. doi:10.1038/sj.onc.1206078 regulation of estradiol production in hormone-depen- dent breast carcinomas (Poutanen et al., 1992; Sasano Keywords: 17q; 17b-HSD; breast cancer; ERBB2 et al., 1996; Suzuki et al., 2000). In a previous study, we estrogen; HSD17B1; real-time PCR found that the expression of both 17b-HSD types 1 and 2 differs in breast cancers with or without late relapse in the disease. Patients with late relapse more often had a Introduction high expression of 17b-HSD type 1 or loss of 17b-HSD type 2 (Gunnarsson et al., 2001). Breast cancer is one of the most common malignancies The gene encoding 17b-HSD type 1 (HSD 17B1)is in women worldwide and is also a major cause of located at 17q12–21, a region that often is rearranged in cancer-related deaths among women. Estrogens play an breast cancer (Plummer et al., 1997; Kauraniemi et al., important role in the development of hormone-depen- 2001). This region also harbors ERBB2, which is dent breast cancer. In postmenopausal women estrogens described as the most amplified gene in breast cancer. in breast carcinoma tissue originate through two main In 20–30% of all breast cancers ERBB2 is over- pathways, one involving aromatase, which converts expressed. High expression is common in ER-negative tumors and gives rise to a worse prognosis (for reviews see Re` villion et al., 1998). In a majority of breast *Correspondence: O Sta˚ l, Department of Oncology, University ERBB2 Hospital, SE-581 85 Linko¨ ping, Sweden; E-mail: [email protected] carcinomas with amplification the neighboring Received 1 August 2002; revised 26 September 2002; accepted 27 topoisomerase IIa gene is either amplified or deleted, September 2002 which in turn has importance for the response to Amplification of HSD17B1 in breast cancer C Gunnarsson et al 35 chemotherapy (Ja¨ rvinen et al., 2000). Furthermore, amplified, whereas samples with levels lower than 0.5 frequent loss of heterozygosity has been observed at were considered to have a decreased copy number. The several loci on 17q, and the tumor suppressor gene mean normalized HSD17B1 level was 1.32 (range 0.2– BRCA1 is located at 17q21 (Bie` che and Lidereau, 1995; 7.9). In all, 32 tumors (14.5%) showed a value higher Osborne and Hamshere, 2000). than 2 and an equal number of tumors showed a value The aim of this study was to investigate if the gene lower than 0.5. The remaining 71% of the tumors had encoding 17b-HSD type 1 exhibits altered gene copy values clustered around the normal level (mean 1.14, number in breast cancer. Using real-time PCR and the 95% CI 1.08–1.21). Figure 1 illustrates a case with and TaqMan methodology, we examined 221 tumors from without amplification, respectively. For ERBB2, the postmenopausal patients for gene amplification of mean value was 2.03 (range 0.3–25.4), with 21% of the HSD17B1 and ERBB2. As a control, we examined the tumors showing gene amplification, 14% showing a low mRNA expression of 17b-HSD type 1 in a subgroup of copy number and 65% showing normal gene levels tumors. (mean 0.96, 95% CI 0.90–1.02). In total, 83% of the tumors with amplification of ERBB2 had a high protein expression compared with 13% of those without gene Results amplification (Po0.00001). Only two out of 32 tumors with a low copy number showed overexpression of the Using the amyloid precursor protein (APP) gene as a protein. A total of 14 out of the 32 tumors (44%) with reference, the HSD17B1 to APP and ERBB2 to APP HSD17B1 amplification showed at the same time copy-number ratios were calculated in 221 tumors. amplification of ERBB2, and the correlation between Samples with values higher than 2 were considered HSD17B1 and ERBB2 amplification was statistically

Figure 1 HSD17B1 and APP gene quantity by real-time PCR in two breast tumor samples. The threshold cycle (Ct) values for APP were almost identical in the two samples, whereas a shift in Ct was evident for HSD17B1, indicating different gene-copy ratios in the two samples, the actual gene-copy ratios were calculated from the standard curves obtained in the same experiment (not shown). The measurements were made in triplicates for each sample

Oncogene Amplification of HSD17B1 in breast cancer C Gunnarsson et al 36 significant (P ¼ 0.00078, Table 1). No correlation was (P ¼ 0.98). In the ER-negative subgroup patients with observed between decreased copy number of HSD17B1 ERBB2 amplification had worse prognosis compared to and altered ERBB2. those without amplification (P ¼ 0.046, Figure 2c). No association was found between HSD17B1 amplification and lymph node status (P ¼ 0.71), tumor size (P ¼ 0.80) or estrogen receptor (ER) status (P ¼ 0.88). Discussion Patients with ERBB2 amplification were to a higher degree ER negative (P ¼ 0.0028), and had larger tumors Estradiol plays a crucial role in breast cancer evolution compared to others (P ¼ 0.016). A total of 48 tumors and it has been shown that in normal breast tissue the were examined with regard to the mRNA expression of oxidative pathway (E2-E1) dominates, whereas in 17b-HSD type 1, among which 16 tumors showed malignant breast tumors the reductive pathway HSD17B1 amplification, 15 tumors had normal gene (E1-E2) is dominant (Speirs et al., 1998a, b). 17b- copy number and 17 tumors showed a low copy number. HSD type 1 is the principal enzyme responsible for the The median (range) expression level in the groups with conversion of estrone (E1) to the more biologically low, normal and increased gene copy number was 0.8 active estradiol (E2) (Luu-Thee et al., 1995). These (0.4–1.9), 1.5 (0.7–14.6), and 2.1 (0.9–8.6), respectively. results suggest that intratumoral regulation of estradiol The difference was statistically significant (P ¼ 0.0010, levels is possible. Kruskal–Wallis test). Dividing the tumors by expression Miyoshi et al. (2001) observed that mRNA levels of level into three categories also showed a significant 17b-HSD type 1, more evident than aromatase and association between gene copy number and expression sulfatase levels, were higher in postmenopausal than level (Table 2). premenopausual tumors, and they suggested that Patients with amplification of either HSD17B1 or upregulation of 17b-HSD type 1 is important in the ERBB2 or both showed lower breast cancer survival than maintenance of high intratumoral E2 levels in post- patients without amplification (P ¼ 0.0059). Both ampli- menopausal patients. On the other hand, Utsumi et al. fications appeared to indicate worse outcome indepen- (1999) suggested steroid sulfatase as a useful prognostic dently (Figure 2a). HSD17B1 turned out to be an marker for the identification of high-risk breast cancer independent prognostic factor with borderline signifi- patients. Feigelson et al. (2001) found that a poly- cance (P ¼ 0.052) in multivariate analysis adjusting for morphism in the gene coding for 17b-HSD type 1 might lymph node status, tumor size, ER status, ERBB2 be used to identify women with an increased risk for amplification and treatment (Table 3). The prognostic advanced breast cancer. In the present study, we found importance of HSD17B1 amplification was greatest for amplification of this gene in 14.5% of the cases, and the ER-positive patients who had received tamoxifen provided that this leads to overexpression of the (P ¼ 0.017, Figure 2b), while there was no significant enzyme, it might be a mechanism of a high E1 to E2 difference for ERBB2 in the same group of patients conversion rate in the tumor. The gene encoding 17b-HSD type 1 is located at 17q12–21 and this region shows frequent genetic Table 1 Amplification of HSD17B1 and ERBB2 for all tumors. The rearrangement (Plummer et al., 1997). The HSD17B1 correlation was statistically significant (P=0.00078) gene originally was a candidate gene for BRCA1 ERBB2 (Mannermaa et al., 1994). After it was eliminated as a Gene copy ratio candidate, it has received little attention despite its putative importance in the synthesis of estradiol. Several r2>2 n (%) n (%) other genes that have been related to breast cancer are located within the same chromosomal bands. Among HSD17B1 those, both ERBB2 and the gene encoding topoisome- Gene copy ratio rase IIa exhibit gene amplification, and recently new r2 156 (71) 33 (15) >2 18 (8) 14 (6) amplified and highly expressed genes were discovered in the 17q12 region (Kauraniemi et al., 2001). We found ERBB2 amplification in 21% of the cases, which falls in the range known from previous studies. We also observed a decreased copy number in 14% of ERBB2. Table 2 Correlation between HSD17B1 gene copy number and expression level of 17b-HSD type 1 (P=0.00005). The expression level Similarly, a low copy number of HSD17B1 was seen in in normal mammary gland was used as a reference (=1) 15% of the tumors, and this correlated with a low expression of mRNA. In the 17q11–21 region not only 17b-HSD type 1 expression level amplifications but also deletions have been described r1 1–2 >2 (Osborne and Hamshere, 2000). Among the tumors with nnngene amplification of HSD17B1, 44% showed coampli- HSD 17B1 fication of ERBB2. Coamplification with ERBB2 has Gene copy ratio earlier been shown in the genes coding for topoisome- o0.5 10 7 0 rase IIa (Ja¨ rvinen et al., 2000), MLN 64 (Bie` che et al., 0.5–2 1 9 5 1996) and placental lactogen, but in the latter case the >2 2 6 8 clinical relevance can be discussed (Latham et al., 2001).

Oncogene Ampliﬁcation of HSD17B1 in breast cancer C Gunnarsson et al 37

Figure 2 (a) Breast cancer survival in relation to amplification of HSD17B1 and ERBB2. The number of patients in the different groups was 156 (ERBB2À/HSD17B1À), 18 (ERBB2À/HSD17B1+), 33 (ERBB2+/HSD17B1À), and 14 (ERBB2+/HSD17B1+), respectively. (b) Breast cancer survival for ER-positive patients treated with tamoxifen in relation to HSD17B1 amplification (P ¼ 0.017). The HSD17B1À and HSD17B1+ groups consisted of 66 and 15 patients, respectively. (c) Breast cancer survival for ER- negative patients in relation to ERBB2 amplification (P ¼ 0.046). The ERBB2À and ERBB2+ groups consisted of 42 and 22 patients, respectively

Oncogene Amplification of HSD17B1 in breast cancer C Gunnarsson et al 38 Table 3 Multivarite analysis of breast cancer mortality rate in estrogen metabolism modulators such as aromatase relation to HSD17B1 amplification and other variables using Cox inhibitors. Interestingly, Ellis et al. (2001) found that the proportional hazard regression relative benefit from levastrol as compared to tamoxifen No. of Rate (95%CI) Significance was significantly higher among ErbB2-positive patients. patients ratio Whether ErbB2 is an independent indicator of tamoxifen Lymph node status resistance is still a question of debate. The present study NÀ 25 1.0 did not demonstrate an interaction between ERBB2 N+ 196 5.2 (2.2–12.2) P=0.00012 amplification and benefit from tamoxifen. However, this Tumor size could be because of the statistical power since the ER+/ r20 mm 86 1.0 ERBB2+ subgroup comprised few patients. 21–30 mm 70 1.4 (0.8–2.3) Po0.00001 >30 mm 65 3.9 (2.3–6.6) Among the tumors without HSD17B1 amplification, some ER status still show a high mRNA expression. Gene amplification is ERÀ 64 1.0 probably not the only reason for overexpression of type 1 and a ER+ 155 0.77 (0.49–1.2) P=0.24 lot of knowledge about the regulation of 17b-HSD type 1 is still ERBB2 status ERBB2À 174 1.0 missing. Speirs et al. (1998a, 1999) suggested that cytokines can ERBB2+ 47 1.4a (0.85–2.3) P=0.18 regulate the E2 synthesis in breast cancer by stimulating enzyme HSD17B1 status activity, they especially pointed out IL-6 and IL-8 as possible HSD17B1 189 1.0 estrogen modulators. Other studies have also shown that IL-6 b b HSD17B1 32 1.7 (1.0–2.9) P=0.052 can upregulate HSD17B1 (Purohit et al., 2002). Adjuvant therapy Radiotherapy 97 1.0 As far as we know, this is the first study reporting on Chemotherapy 124 1.05 (0.70–1.6) P=0.82 amplification of the gene coding for 17b-HSD type 1. Endocrine treatment This event may lead to higher levels of estradiol in the No tamoxifen 104 1.0 tumor. We found that amplification of the gene had Tamoxifen 117 0.76 (0.51–1.13) P=0.18 prognostic significance. In particular, an increased gene aModel excluding HSD17B1; RR=1.6, P=0.043 copy number indicated a decreased breast cancer bModel excluding ERBB2; RR=1.9, P=0.012 survival for ER-positive patients who received tamoxifen treatment.

The position of HSD17B1 appears not to be close to the Materials and methods center of the ERBB2 amplicon and the gene might belong to a neighboring amplicon within 17q12–21. We Patients and tumors have estimated the distance between the two genes to be The postmenopausal patients in this study were randomized to 2.9 Mb (NCBI Map Viewer). The gain in gene copy adjuvant chemotherapy or postoperative radiotherapy. The number was moderate in most of the amplified tumors patients had either histologically verified lymph node metas- as was the overexpression of the gene. On the other tases or a tumor diameter exceeding 30 mm. The patients did hand, the clinical outcome was significantly worse not receive any preoperative treatment. Surgery consisted of among patients with HSD17B1 amplification, which modified radical mastectomy. For most of the patients was further supported by the multivariate analysis randomized to chemotherapy, the treatment consisted of 12 (Table 3). In particular, ER-positive patients who courses of CMF given according to the original Milan protocol (cyclophosphamide 100 mg/m2 orally on days 1–14, received tamoxifen, a group with good prognosis in metotrexate 40 mg/m2 i.v. on days 1 and 8, and 5-fluorouracil general, showed decreased breast cancer survival if the 600 mg/m2 i.v. on days 1 and 8). Radiotherapy was given with tumor showed amplification of HSD17B1 (P ¼ 0.017). the high-voltage technique with a total dose of 46 Gy. Using a Still, the observed correlation between HSD17B1 2 Â 2 factorial study design, the patients were also randomized amplification and poor prognosis might just be a marker between tamoxifen or no adjuvant endocrine therapy. Tamox- for another nearby gene. In a previous immunohisto- ifen treatment was given postoperatively at a dose of 40 mg chemical study no correlation was found between 17b- daily for at least 2 years. The current study included a subset HSD type 1 and prognosis (Suzuki et al., 2000). consisting of 221 patients from whom frozen tumor tissue was However, this study consisted of a relatively small still available after hormone receptor assays. This number did number of patients and no adjustment was made for ER not include samples that showed a low percentage of malignant cells (o50%) as judged from a 5 mm tissue section status. We have previously found that high mRNA adjacent to the 25 mm section used for DNA extraction. The expression of 17b-HSD type 1 or low mRNA expression patients were followed for a median period of 11.3 years, and of 17b-HSD type 2 predicts recurrence among ER- deaths due to breast cancer were registered in 102 cases during positive patients (Gunnarsson et al., 2001). the follow-up period. ER content was measured with Some ER-positive patients with metastatic disease immunoassays (Abbott Laboratories, Chicago, IL, USA). who have failed on tamoxifen still respond to aromatase Samples with >0.1 fmol/mg DNA were classified as positive. inhibitors (Dowsett, 1997). In the presence of high intratumoral estradiol levels, tamoxifen may not fully Protein expression block the action of estrogen. This could point out Overexpression of the ErbB2 protein was quantified with an amplification of HSD17B1 as a possible event in immunocytochemical method using flow cytometry (Sta˚ l et al., tamoxifen resistance, and future studies may show 1994). Briefly, a cell suspension was prepared from frozen whether it could be a marker for treatment with selective tissue and fixed in 1% PFA for 3 min. The cells were incubated

Oncogene Amplification of HSD17B1 in breast cancer C Gunnarsson et al 39 with a monoclonal antibody (c-neu, Ab-2, Oncogene Research TGGG C CGCACT-30 (reverse), 50-CGATCAGGCTCAAGT Products, Cambridge, MA, USA), or an irrelevant isotype GGACCCCAA-30 (probe). We conducted Blast searches control antibody, followed by incubation with a secondary (Genbank) to confirm the specificity of nucleotide sequences FITC-conjugated antibody. A measure of erbB2 expression chosen for the primers and probes and the absence of DNA was calculated from the paired samples. The cut off value for polymorphism. To avoid detection of contaminating overexpression was the same as used in previous studies (Sta˚ l genomic DNA, the probe was placed at the junction between et al., 2000). two exons.

DNA extraction Real-time polymerase chain reaction Tumor tissue was disintegrated with scissors and then digested All reactions were performed in the ABI Prism 7700 Sequence with 50 ml proteinase K (10 mg/ml, Boehringer Mannheim) in Detection System (Applied Biosystems) using TaqMan meth- 500 ml TEN buffer (10 mm Tris, 1 mm EDTA, 100 mm NaC1) odology. Reactions are characterized by the point during containing 100 ml SDS (20%) and incubated at 551C. DNA cycling when amplification of the PCR product is first was separated from proteins by repeated extractions with detected, rather than the amount of PCR product after a fixed phenol, phenol/chloroform (1 : 1) and chloroform. Nucleic number of cycles. The design of the TaqMan probes, combined acids were precipitated by addition of two volumes of ice-cold with the 50–30 nuclease activity of AmpliTaqGold DNA 95% ethanol and 1/10 volume 3 m sodium acetate incubated polymerase (Applied Biosystems), allows the direct detection overnight at À201C. DNA was pelleted by centrifugation at of the PCR product by the release of a fluorescent reporter 15 000 Â g for 1 h, after which salt residues and RNA were during the PCR. We examined the DNA content of HSD17B1 removed by washing with 70% ethanol. Nucleic acids were and ERBB2 using APP as a control gene in the 221 tumors. repelleted as above for 10 min and then vacuum-dried and The breast cancer cell line MDA-MB-231 was used to suspended in sterile Milli-Q water. construct standard curves. MDA-MB-231 was purchased from American type culture collection (ATCC). MDA-MB-231 RNA extraction cells show normal gene copy levels at 21q21 and 17q12–21 (http://www.nhgri.nih.gov/DIR/CGB/CR2000/). We also Frozen breast tumor tissue (30 mg), was homogenized in a investigated the mRNA level of 17b-HSD type 1 in 48 of microdismembrator (B. Brown, Melsungen, Germany) and the the tumors, using the expression of 17b-HSD type 1 in total RNA was extracted using the SV Total RNA Isolation normal mammary gland as a reference. Total RNA from System (Promega, SDS, Sweden). The purified RNA was mammary gland was purchased from Clontech (Becton stored at À701C and the concentration of RNA content was Dickinson, Sweden). determined by spectrophotometry. cDNA synthesis PCR conditions In all, 400 ng of total RNA was reverse transcribed in a final cDNA (3 ml) or DNA was added to the reaction mixture that volume of 20 ml using the GIBCO BRL kit (Invitrogen, had a total volume of 25 ml. Using TaqMan PCR core Reagent Sweden) with the following concentrations: 1 Â first strand kit (Applied Biosystems), the concentrations used were: 1x buffer, 0.5 mm dNTP, 2.5 mm random hexamers, 10 mm DTT, TaqMan buffer A, 5.0 mm MgCl, 0.1 mm dNTP, 0.1 mm each 0.5 ml Superscript II reverse transcriptase (Invitrogen, Sweden). of forward and reverse primer, 0.1 mm probe and 0.025 U/ ml The thermal conditions used were: 201C for 10 min, 421C for AmpliTaqGold DNA polymerase. The thermal conditions 50 min, 991C for 5 min, and 51C for 5 min. The samples were used were 951C for 10 min, 951C for 15 s and 601C for 1 min. stored at 41C until analyzed. Steps two and three were repeated for 50 cycles. The different targets were amplified independently in separate reaction wells, in triplicates. On the same plate we included samples for Primers and probes standard curves for the target genes. Primers and probes were designed to recognize human HSD17B1, ERBB2 and APP (amyloid precursor protein), Standard-curve method which was used as reference gene. The location of this gene at 21q21 is not known to show chromosomal alterations in breast A relative kinetic method was applied using a standard curve. cancer (Bie` che et al., 1998; Osborne and Hamshere, 2000). The This was constructed with four fold serial dilutions of DNA primer and probe sequences used were as follows: HSD17B1: from the breast cancer cell line MDA-MB-231 or of cDNA 50-CGCCTCAGGAACCTCGTCT-30 (forward), 50-GCTG- from normal human breast tissue. The target content in GCGCAATAAACGTCA-30 (reverse), 50-CCCCACCTAAG- unknown samples was quantified by using the standard curves GGCTGCCTTTCAA-30 (probe); ERBB2:50-GGTCCTGG to determine a relative measure of the starting amount. Each AAGCCACAAGG-30 (forward), 50-GGTTTTCCCACCA- sample was then normalized on the basis of its APP content CATCCTCT-30 (reverse), 50-AACACAACACATC CCCCTC- (DNA) of b-actin content (cDNA). CTTGACTATCAA-30 (probe); and APP:50-TTTGTGTGC- TCTCCCAGGTCT-30 (forward), 50-TGGTCACTGGTTG- Statistical analysis GTTGGC-30 (reverse), 50-CCCTGAACTGCAGATCAC- CAATGTGGTAG-30 (probe). Both primers and probes were The relationships between grouped variables were analyzed purchased from Applied Biosystems. with the w2 test, or the w2 test for the trend when required. Primers and probe were designed to recognize human 17b- Survival curves were produced according to the life-table HSD type 1 cDNA sequences using the computer program method described by Kaplan and Meier. Differences in breast Primer Express (Applied Biosystems) and b-actin was used as cancer mortality rates were estimated with Gehan’s test. endrogenous control (Applied Biosystems). The primer and Multivariate analysis of mortality rates was performed with probe sequences for 17b-HSD type 1 were as follows: 50- Cox proportional hazard regression. All the procedures are TATGCGAGAGTCTGGCGGTT-30 (forward),50-TGCAC comprised in the statistical package STATISTICA 6.0

Oncogene Ampliﬁcation of HSD17B1 in breast cancer C Gunnarsson et al 40 (StatSoft Scandinavia AB, Sweden). The criterion for statis- Acknowledgements tical signiﬁcance was Po0.05. This study was supported by grants from the Swedish Cancer Society.

References

Bie` che I and Lidereau R. (1995). Genes Chrom. Cancer, 14, Osborne RJ and Hamshere MG. (2000). Cancer Res., 60, 227–251. 3706–3712. Bie` che I, Olivi M, Champe` me M-H, Vidaud D, Lidereau R Plummer S, Paris M, Myles J, Tubbs R, Crowe J and Casey G. and Vidaud M. (1998). Int J. Cancer, 78, 661–668. (1997). Genes Chrom. Cancer, 20, 354–362. Bie` che I, Tomasetto C, Regnier CH, Moog-Lutz C, Rio MC Poutanen M, Isomaa V, Letho V-P and Vihko R. (1992). Int. and Lidereau R. (1996). Cancer Res., 56, 3886–3890. J. Cancer, 50, 386–390. Dowsett M. (1997). J. Steroid Biochem. Mol. Biol., 61, 261– Purohit A, Newman SP and Reed MJ. (2002). Breast Cancer 266. Res., 4, 65–69. Ellis M, Coop A and Singh B. (2001). J. Clin. Oncol. 19, 3808– Re` villion F, Bonneterre J and Peyrat JP. (1998). Eur. J. 3816. Cancer, 34, 791–808. Feigelson SH, McKean-Cowdin R, Coetzee GA, Stram DO, Sasano H, Frost AR, Saitoh R, Harada N, Poutanen M, Kolonel LN and Henderson BE. (2001). Cancer Res., 61, Vihko R, Buhm SE, Silverberg SG and Nagura H. (1996). J. 785–789. Clin. Endocrin. Metabol., 11, 4042–4046. Gunnarsson C, Olsson B and Sta˚ l O. (2001). Cancer Res., 61, Speirs V, Green AR and Atkin SL. (1998a). J. Steroid 8448–8451. Biochem. Mol. Biol., 67, 267–274. Ja¨ rvinen TA, Tanner M, Rantanen V, Barlund M, Borg A˚ , Speirs V, Green AR, Walton DS, Kerin MJ, Fox JN, Carleton Grenman S and Isola J. (2000). Am. J. Pathol., 156, PJ, Desai SB and Atkin SL. (1998b). Br. J. Cancer, 78, 839–847. 1421–1429. Kauraniemi P, Barlund M, Monni O and Kallioniemi A. Speirs V, Walton DS, Hall M-C and Atkin SL. (1999). Br. J. (2001). Cancer Res., 61, 8235–8240. Cancer, 81, 690–695. Labrie F, Luu-Thee V, Lin S-X, Labrie C, Simard J, Breton R Sta˚ l O, Sullivan S, Sun X-F, Wingern S and Nordenskjo¨ ld B. and Belanger A. (1997). Steroids, 62, 148–158. (1994). Cytometry, 16, 160–168. Latham C, Zhang A, Nalbanti A, Maner S, Zickert P, Belgen Sta˚ l O, Borg A˚ , Ferno¨ M, Ka¨ llstro¨ m A-C, Malmstro¨ mP, H and Zetterberg A. (2001). Cancer Genet. Cytogenet., 127, Nordenskjo¨ ld B and members of the South Sweden Breast 16–23. Cancer Group and the Southeast Sweden Breast Cancer Luu-Thee V, Zhang Y, Poirier D and Labrie F. (1995). Group. (2000). Ann. Oncol., 11, 1545–1550. Biochem. Mol. Biol., 55, 581–587. Suzuki T, Moriya T, Ariga N, Kanazawa M and Sasano H. Mannermaa A, Peltoketo H, Winqvist R, Ponder BA, (2000). Br. J. Cancer, 82, 518–523. Kiviniemi H, Easton DF, Poutanen M, Isomaa V and Utsumi T, Yoshimura N, Takeuchi S, Ando J, Maruta M, Vihko R. (1994). Human Genet., 93, 319–324. Maeda K and Harada N. (1999). Cancer Res., 59, 377–381. Miettinen M, Mustonen M, Poutanen M, Isomaa V and Vihko Vihko P, Isomaa V and Ghosh D. (2001). Mol. Cell. Endocrin., R. (1996). J. Biochem., 314, 839–845. 171, 71–76. Miyoshi Y, Ando A, Shiba E, Taguchi T, Tamaki Y and Yue W, Wang J, Hamilton C, Demers L and Sauten R. (1998). Noguchi S. (2001). Int. J. Cancer, 94, 685–689. Cancer Res., 58, 927–932.

Oncogene