HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Supplementary data Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Tozlu-Kara, S Articles by Bièche, I Search for Related Content PubMed PubMed Citation Articles by Tozlu-Kara, S Articles by Bièche, I

¹ Centre René Huguenin,, FNCLCC, F-92210 St-Cloud, France ² INSERM,, U735, F-92210 St-Cloud, France ³ Functional Genomic Unit,, Institut Gustave-Roussy, Villejuif, France ⁴ ISPB,, Faculté de Pharmacie de Lyon, Université Claude Bernard Lyon 1, Lyon F-69008, France ⁵ Inserm,, U590, Lyon F-69008, France ⁶ Centre Léon Bérard,, Lyon F-69008, France ⁷ Département de Médecine,, Centre René Huguenin, St-Cloud, France ⁸ CNRS UMR 8125 Bioinformatics Unit,, Institut Gustave-Roussy, Villejuif, France

(Correspondence should be addressed to I Bièche; Email: i.bieche{at}stcloud-huguenin.org)

	Abstract

Top Abstract Introduction Materials and methods Results Discussion References

 
The estrogen receptor (ER) status of breast tumors is used to identify patients who may respond to endocrine agents such as tamoxifen. However, ER status alone is not perfectly predictive, and there is a pressing need for more reliable markers of endocrine responsiveness. In this aim, we used a two-step strategy. We first screened genes of interest by a pangenomic 44 K oligonucleotide microarray in a series of ten ER-positive tumors from five tamoxifen-treated postmenopausal patients who relapsed (distant metastasis) and five tamoxifen-treated postmenopausal patients who did not relapse, matched with respect to age, Scarff–Bloom–Richardson grade, lymph node status, and macroscopic tumor size. Genes of interest (n=24) were then investigated in an independent well-characterized series of ER-positive unilateral invasive primary breast tumors from postmenopausal women who received tamoxifen alone as adjuvant hormone therapy after primary surgery. We identified four genes (HRPAP20, TIMELESS, PTPLB, and MGC29814) for which high mRNA levels were significantly associated with shorter relapse-free survival (log-rank test). We also showed that hormone-regulated proliferation-associated 20 kDa protein (HRPAP20) and TIMELESS are 17ß-estradiol-regulated in vitro and are ectopically expressed in OH-Tam-resistant cell lines. In conclusion, these findings point to HRPAP20 and TIMELESS as promising markers of tamoxifen resistance in women with ER-positive breast tumors.

	Introduction

Top Abstract Introduction Materials and methods Results Discussion References

 
Tamoxifen is by far the most common endocrine agent used to treat women with all stages of breast cancer, and particularly for postmenopausal patients. Resistance to tamoxifen is a major problem. Estrogen receptor (ER) expression is used to identify breast cancer patients who are likely to respond to tamoxifen, but half of all patients with ER-positive tumors fail to respond favorably to antiestrogen treatment (McGuire 1980). Moreover, the proportion of ER-positive tumors is higher in postmenopausal patients than in younger patients, meaning that ER status is poorly predictive in this population (Diab et al. 2000). Combined assessment of ER and ER-inducible genes (as markers of functional ER-mediated cell growth mechanisms) might be more reliable. Measurement of tumor progesterone receptor and PS2 expression is slightly more reliable than ER assay alone in this setting. However, the expression of PR and PS2 correlates strongly with ER expression, meaning that the PS2 and PR markers provide little further information on hormone dependence relative to ER. There is thus a pressing need for more reliable markers of endocrine responsiveness.

The advent of efficient tools for large-scale gene expression analysis has provided new insights into the involvement of gene networks and regulatory pathways in various tumoral processes (DeRisi et al. 1996). cDNA and oligonucleotide microarrays are used to test the expression pattern of thousands of genes at a time in individual tumors, and to identify genes or molecular signatures that are predictive of outcome (van't Veer et al. 2002).

Here, to identify candidate markers with which to predict the response to tamoxifen, we first used pangenomic dual color 44 K long oligonucleotide microarrays to screen ER-positive tumors from ten postmenopausal breast cancer patients treated with adjuvant tamoxifen alone, of whom five relapsed (distant metastasis). Twenty-four genes of interest thus identified were then investigated in an independent panel of ER-positive tumors from 48 tamoxifen-treated postmenopausal breast cancer patients, of whom 24 relapsed. Overexpression of four genes (HRPAP20, TIMELESS, PTPLB, and MGC29814) was associated with significantly shorter relapse-free survival in univariate analysis. We also found that HRPAP20 and TIMELESS expression patterns differed between a hydroxy (OH)-Tam-sensitive cell line (MVLN) and two MVLN-derived OH-Tam-resistant cell lines.

	Materials and methods

Top Abstract Introduction Materials and methods Results Discussion References

 
Patients and samples

We analyzed samples of 58 primary breast tumors excised from women at our institution from 1980 to 1994. Samples containing more than 70% of tumor cells were considered suitable for this study. Immediately following surgery, the tumor samples were placed in liquid nitrogen until RNA extraction. Ten tumors were used for oligonucleotide microarray analysis (screening set) and another 48 tumors were used for real-time quantitative RT-PCR (validation set).

The patients met the following criteria: primary unilateral non-metastatic postmenopausal breast carcinoma; ER positivity (as determined at the protein level by biochemical methods (dextran-coated charcoal method until 1988 and enzymatic immunoassay thereafter) and confirmed by ER real-time quantitative RT-PCR assay); complete clinical, histological, and biological information available; no radiotherapy or chemotherapy before surgery; and full follow-up at our institution. Standard prognostic factors are shown in Tables 1 and 2. The patients had physical examinations and routine chest radiography every 3 months for 2 years, then annually. Mammograms were done annually. The median follow-up was 7.2 years (range 1.5–10.0 years). All the patients received postoperative adjuvant endocrine therapy (20 mg tamoxifen daily for 3–5 years), and no other treatment. Twenty-nine patients relapsed. The distribution of first relapse events was as follows: 26 metastases and 3 local and/or regional recurrences with metastases.

View this table: [in this window] [in a new window]   Table 1 Characteristics of the ten estrogen receptor (ER)-positive tumors used in oligonucleotide microarray analysis (matched pairwise with respect to age, Scarff–Bloom–Richardson (SBR) grade, lymph node status, and macroscopic tumor size), from patients with and without relapse

View this table: [in this window] [in a new window]   Table 2 Characteristics of the 48 estrogen receptor (ER)-positive tumors from patients with and without relapse, used for real-time RT-PCR analysis

Specimens of adjacent normal breast tissue from four breast cancer patients and normal breast tissue from three women undergoing cosmetic breast surgery were used as sources of normal RNA.

For the analysis of genes of interest in normal adult human tissues, total RNAs from a variety of normal adult human tissues were purchased from Clontech. The tissues analyzed were breast, placenta, liver, spleen, testis, ovary, kidney, lung, trachea, heart, bone marrow, skeletal muscle, small intestine, thyroid, colon, thymus, prostate, uterus, salivary gland, leukocytes, bladder, spinal cord, adrenal gland, and brain.

Oligonucleotide microarray

We analyzed ten ER-positive breast tumors, consisting of five tumors with relapses (distant metastasis) matched to five tumors without relapse with respect to age, Scarff–Bloom–Richardson (SBR) grade, lymph node status, and macroscopic tumor size (Table 1).

The quality of the ten ER-positive breast tumor total RNAs, based on the 28S/18S rRNA ratio, was assessed by the RNA 6000 Nano Lab-On-chip on the Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA). All specimens had a 28S-to-18S ratio higher than 1.5. A pool composed of equal amounts of total RNA from the ten ER-positive breast tumors was used as the RNA reference. Five hundred nanograms of total RNA from each sample and from the RNA reference pool were used to generate labeled antisense cRNAs with T7 RNA polymerase.

We used the Agilent 44 K Whole Human Genome (G4112A) long (60 bp) oligonucleotide microarray and the dual-color analysis method in which probes from tumor samples and from reference RNA are differentially labeled with cyanine 5 and cyanine 3. These microarrays developed by Agilent Technologies have 44 290 features with 41 000 distinct oligonucleotides belonging to 33 715 sequences defined by their accession number. Due to the redundancy of probes and accession number, they represent around 18 795 symbols of Entrez_Gene (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=gene).

Oligonucleotide microarray experiments were performed with duplicates for each data point, using a dye-swap design to avoid dye bias (20 oligonucleotide microarray experiments in total): cRNA from each tumor was labeled with cyanine 3 (Cy3)-cytidine triphosphate (CTP) and the RNA reference pool with cyanine 5 (Cy5)-CTP for the direct comparison and vice versa for the dye-swap analysis.

Reverse transcription, linear amplification, cRNA labeling, and purification were performed with the Agilent linear amplification kit. The cRNA concentration and Cy3-CTP or Cy5-CTP incorporation were assessed with a u.v.–visible spectrophotometer. Hybridization was run for 17 h at 60 °C, with 1 µg Cy3- or Cy5-labeled cRNA from each tumor, mixed with the same amount of Cy5- or Cy3-labeled cRNA from the reference pool. The arrays were then washed with 0.6x and 0.01x SSC buffers containing Triton, and dried with a nitrogen gun. Microarrays were scanned with the Agilent DNA Microarray Scanner.

Feature Extraction software (Agilent Technologies) was used to quantify the intensity of fluorescent images and to normalize results by the local background subtraction option. Files used for statistical analysis contained, for each tumor sample, the list of 44 000 features associated with a set of values including the log ratio compared with the reference, the P value of the log ratio, and intensities. Dye-swap designs, in which each tumor is processed in duplicate, were used to assess reproducibility, and improved both the efficiency and the validity of the data. The final ratio (i.e., the fold change) is the average of the two individual ratios from the dye-swap experiment (combined experiments).

All data obtained by microarray analysis have been submitted to Array Express at the European Bioinformatics Institute (http://www.ebi.ac.uk/arrayexpress/) with accession number E-TABM-72. ArrayExpress (at the European Bioinformatics Institute) is a public repository for microarray data, which is aimed at storing well-annotated data in accordance with Microarray Gene Expression Data recommendations (http://www.mged.org).

Microarray data were mainly analyzed with Resolver software (Rosetta Inpharmatics, Seattle, WA, USA). All data were filtered to eliminate low-intensity values under 200 arbitrary units for both colors, a threshold determined on the basis of the linearity test. Each selected gene had at least a twofold change, with a P value <0.001. Using this procedure, 5234 genes passed the filter. For unsupervised clustering, we used a hierarchical agglomerative algorithm that pairs samples according to a Pearson-based distance.

ANOVA was applied to the microarray data in order to discriminate between breast tumors with and without relapse.

Real-time RT-PCR

Theoretical basis
Reactions are characterized by the point during cycling when amplification of the PCR product is first detected, rather than the amount of PCR product accumulated after a fixed number of cycles. The larger the starting quantity of the target molecule, the earlier a significant increase in fluorescence is observed. The parameter threshold cycle (C_t) is defined as the fractional cycle number at which the fluorescence generated by SYBR green dye-amplicon complex formation passes a fixed threshold above baseline. The increase in the fluorescence signal associated with exponential growth of PCR products is detected by the laser detector of the ABI Prism 7700 Sequence Detection System (Perkin–Elmer Applied Biosystems, Foster City, CA, USA), using PE Biosystems analysis software according to manufacturer's manuals.

The precise amount of total RNA added to each reaction mix (based on optical density) and its quality (i.e., lack of extensive degradation) are both difficult to assess. We therefore also quantified transcripts of two endogenous RNA control genes involved in two cellular metabolic pathways, namely TATA-binding protein (TBP; Genbank accession NM_003194 [GenBank] , which encodes the TATA box-binding protein (a component of the DNA-binding protein complex transcription factor IID (TFIID), and RPLP0 (also known as 36B4; NM_001002 [GenBank] , which encodes human acidic ribosomal phosphoprotein P0. Each sample was normalized on the basis of its TBP (or RPLPO) content.

Results, expressed as N-fold differences in target gene expression relative to the TBP (or RPLPO) gene, and termed ‘Ntarget’, were determined as Ntarget=2^C_tsample, where the C_t value of the sample is determined by subtracting the average C_t value of the target gene from the average C_t value of the TBP (or RPLP0) gene (Bieche et al. 1999, 2001).

Primers and controls
Primers for TBP, RPLP0, and the target genes were chosen with the assistance of the Oligo 5.0 computer program (National Biosciences, Plymouth, MN, USA).

We conducted searches in the dbEST and nr databases to confirm the total gene specificity of the nucleotide sequences chosen as primers, and the absence of single nucleotide polymorphisms. In particular, the primer pairs were selected to be unique relative to the sequences of closely related family member genes or of the corresponding retropseudogenes. To avoid amplification of contaminating genomic DNA, one of the two primers was placed at the junction between two exons. In general, amplicons were between 70 and 120 nucleotides long. Gel electrophoresis was used to verify the specificity of PCR amplicons.

For each primer pair, we performed no-template control (NTC) and no-reverse transcriptase control (RT-negative) assays, which produced negligible signals (usually >40 in C_t values), suggesting that primer–dimer formation and genomic DNA contamination effects were negligible.

RNA extraction, cDNA synthesis, and PCR conditions have been described elsewhere (Bieche et al. 2001).

Statistical analysis
As the mRNA levels did not fit a Gaussian distribution, a) the mRNA levels in each subgroup of samples were characterized by their median values and ranges, rather than their mean values and coefficients of variation and b) relationships between the molecular markers and clinical and biological parameters were tested by the non-parametric Mann–Whitney U test (Mann & Whitney 1947). Differences between two populations were judged significant at confidence levels >95% (P<0.05).

To visualize the capacity of a given molecular marker to discriminate between two populations (in the absence of an arbitrary cutoff value), we summarized the data in a receiver operating characteristics (ROCs) curve (Hanley & McNeil 1982). These curves plot sensitivity (true-positives) on the y-axis against one-specificity (false-positives) on the x-axis, considering each value as a possible cutoff. The area under curve (AUC) was calculated as a single measure of the discriminatory capacity of each molecular marker. When a molecular marker has no discriminative value, the ROC curve lies close to the diagonal and the AUC is close to 0.5. When a marker has strong discriminative value, the ROC curve moves to the upper left-hand corner (or to the lower right-hand corner) and the AUC is close to 1.0 (or 0).

Relapse-free survival (RFS) was determined as the interval between diagnosis and detection of the first relapse (local and/or regional recurrence, and/or distant metastasis). Survival distributions were estimated by the Kaplan–Meier method (Kaplan & Meier 1958), and the significance of differences between survival rates was ascertained by the log-rank test (Peto et al. 1977).

Cell culture

We studied the following three human breast carcinoma cell lines in various pharmacological conditions: an OH-Tam-sensitive cell line (MVLN) and two MVLN-derived OH-Tam-resistant cell lines (CL6.8 and CL6.32). MVLN is an ER-positive and hormone responsive human breast carcinoma cell line derived from MCF-7 (Demirpence et al. 1993). Treatment of MVLN cells for 6 months led to the emergence of the OH-Tam-resistant (but still estrogen-dependent) individual cellular clones CL6.8 and CL6.32 (Badia et al. 2000) which have been characterized in a previous work (Vendrell et al. 2005).

MVLN, CL6.8, and CL6.32 cells were grown for one or two passages as previously described (Demirpence et al. 1993), then purged for 4 days in Dulbecco's modified Eagle medium without phenol red, supplemented with 3% of steroid-depleted, dextran-coated charcoal-treated fetal calf serum (DCC medium). Cells were then treated for 4 days (with one medium change) under the following pharmacological conditions: steroid-depleted medium (vehicle), 1 nM E₂ (17ß-estradiol), 200 nM OH-Tam, or both 1 nM E₂ and 200 nM OH-Tam.

	Results

Top Abstract Introduction Materials and methods Results Discussion References

 
Screening of differentially expressed genes

Microarray experiments based on dual-color technology were performed with ten ER-positive breast tumor RNAs (five tumors with relapse and five tumors without relapse), relative to an RNA reference pool prepared by mixing identical amounts of RNA from the same ten breast tumors. Each breast tumor sample was analyzed in duplicate using a dye-swap design (20 oligonucleotide microarray experiments in total).

Using Resolver software, an intensity- and fold change-based filtering approach (greater than twofold change and P value <0.001) selected 5234 features from the 44 290 present on the 60-mer oligonucleotide microarray. The unsupervised hierarchical clustering of these 5234 features is presented in Fig. 1. The duplicates (dye-swap) of each breast tumor sample clustered tightly together. On this unsupervised hierarchy, the samples fell into two subgroups: one contained four breast tumors, of which one (25%) was associated with relapse, while the second contained the other six breast tumors, four (67%) of which were associated with relapse.

View larger version (8K): [in this window] [in a new window]   Figure 1 Unsupervised hierarchical clustering of ten ER-positive breast tumors (T1–T10) tested in duplicate with a dye-swap design (20 oligonucleotide microarray experiments in total). The classification was determined by the 5234 filtered features, with an agglomerative algorithm based on the Pearson correlation coefficient. R+, relapse; R–, no relapse; and -dw, dye-swap replicate.

View this table: [in this window] [in a new window]   Table 3 Genes perfectly discriminated between five estrogen receptor (ER)-positive breast tumors with relapse and five breast tumors without relapse

mRNA expression of ESR1/ER, MKI67, and the 24 candidate genes in 24 patients who relapsed and 24 patients who did not relapse

The expression level of the 24 genes identified by oligonucleotide microarray screening and discriminating perfectly between breast tumors with and without relapse was then determined by real-time RT-PCR in an independent cohort of 24 ER-positive breast tumor patients who relapsed and 24 ER-positive breast tumor patients who did not relapse (Table 2). All these 48 ER-positive tumors were from postmenopausal patients treated with primary surgery followed by adjuvant tamoxifen alone. Six (46%) of the 13 upregulated genes identified by oligonucleotide microarray analysis were significantly upregulated in the 24 individual ER-positive breast tumors with relapse, as compared with the 24 ER-positive breast tumors without relapse (P<0.05; Table 4). The six genes significantly upregulated were HRPAP20, TIMELESS, NEK2, PTPLB, MGC29814, and RANBP2L1 (Table 4). The capacity of each of these six genes to discriminate between breast tumors with and without relapse was then tested by ROC curve analysis. The overall diagnostic values of the six molecular markers were assessed in terms of their AUC values (Table 4). HRPAP20 emerged as the most discriminatory marker of relapse status (ROC–AUC, 0.911).

View this table: [in this window] [in a new window]   Table 4 Statistical analysis of mRNA expression of genes in 24 estrogen receptor (ER)-positive breast tumors with relapse relative to 24 ER-positive breast tumors without relapse

ESR1/ER

None of the 11 downregulated genes identified by oligonucleotide microarray analysis was significantly downregulated in the 24 breast tumors with relapse as compared with the 24 tumors without relapse.

The mRNA levels indicated in Table 4 (calculated as described in Materials and methods) show the abundance of the target relative to the endogenous control (TBP) used to normalize the starting amount and quality of total RNA. Similar results were obtained with a second endogenous control, RPLP0 (also known as 36B4). Indeed, the six upregulated genes were also significantly upregulated in tumor samples from the patients who relapsed relative to the patients who did not relapse.

Prognostic value of HRPAP20, NEK2, TIMELESS, PTPLB, MGC29814, and RANBP2L1

We used univariate analysis (log-rank test) to further study the prognostic value of HRPAP20, NEK2, TIMELESS, PTPLB, MGC29814, and RANBP2L1. For each gene, the 48 ER-positive breast tumors were divided into two groups of 24 tumors with ‘low’ and ‘high’ mRNA levels. Univariate analysis showed that a high expression level of HRPAP20, TIMELESS, PTPLB, and MGC29814 correlated with significantly shorter RFS (Fig. 2A–D respectively). The outcomes of the 24 patients with high mRNA levels of these four genes were significantly worse than those of the 24 patients with low mRNA levels. No significant prognostic value was associated with the two other genes, NEK and RANBP2L1 (Fig. 2E and F respectively).

View larger version (16K): [in this window] [in a new window]   Figure 2 RFS curves for tumors with low target gene expression (n=24) and high target gene expression (n=24).

To further investigate the potential endocrine responsiveness function of HRPAP20, TIMELESS, PTPLB, and MGC29814 in ER-positive breast cancer, we analyzed mRNA levels of these four genes by real-time RT-PCR in MVLN, CL6.8, and CL6.32 cells after 4 days of E₂ and/or OH-Tam treatment.

mRNA expression of PTPLB and MGC29814 did not vary in any of the cells lines during any of the treatments.

Figure 3 illustrates compiled expression data for the other two genes (HRPAP20 and TIMELESS), as well as the well-known ER-induced gene pS2/TFF1, in the three cell lines after pharmacological treatment. As pS2/TFF1, HRPAP20, and TIMELESS responded positively to E₂ in the three cell lines (greater than twofold upregulation).

View larger version (18K): [in this window] [in a new window]   Figure 3 mRNA levels of HRPAP20, TIMELESS, and pS2/TFF1 in MVLN, CL6.8, and CL6.32 cells after 4 days of culture in steroid-depleted medium (vehicle), or with E2 and/or OH-Tam. For each gene and each cell line, mRNA levels were normalized such that the value of the vehicle sample was 1. The mRNA levels indicated are means from at least three independent experiments.

Interestingly, the reverse pharmacology of OH-Tam on these two genes was lost in the two OH-Tam-resistant cell lines CL6.8 and CL6.32. Indeed, in the CL6.8 or CL6.32 cells, a new OH-Tam response emerged, characterize either by no activity on HRPAP20 or TIMELESS gene expression under OH-Tam treatment alone (HRPAP20 gene expression in CL6.8 and CL6.32 cell and TIMELESS gene expression in CL6.32 cells; Fig. 3) or by the transformation of the reverse pharmacological action of OH-Tam in an agonist (‘estrogen-like’) activity (TIMELESS gene expression in the CL6.8 cells; Fig. 3). Finally, when the cell lines were treated with both E₂ and OH-Tam, OH-Tam counteracted the E₂-induced upregulation of HRPAP20 and TIMELESS in MVLN cells but not in the two OH-Tam-resistant cell lines, CL6.8 and CL6.32.

To further know whether HRPAP20 and TIMELESS genes are under direct or indirect regulation of the ER, we analyzed HRPAP20, TIMELESS, and pS2/TTF1 gene expression in the MVLN cell line treated for 4 days, but also following two shorter E₂ treatments, i.e., 6 and 18 h. Figure 4 shows that HRPAP20 and TIMELESS genes responded early (from 6 h) to E₂ in the MVLN cell line, these results suggest that, as pS2/TTF1, HRPAP20, and TIMELESS genes are regulated directly by estrogen.

View larger version (16K): [in this window] [in a new window]   Figure 4 mRNA levels of HRPAP20, TIMELESS, and pS2/TFF1 in MVLN after 8 h, 18 h, and 4 days of culture in steroid-depleted medium (vehicle) or with E2. For each gene, mRNA levels of each E2-treated sample were normalized such that the value of the vehicle sample was 1. The mRNA levels indicated are means from at least three independent experiments.

mRNA expression of HRPAP20 and TIMELESS in breast tumor cell lines

The expression level of HRPAP20 and TIMELESS was then determined in nine well-characterized breast tumor cell lines, including four ER-positive cell lines (MCF-7, T-47D, ZR-75-1, and MDA-MB361) and five ER-negative cell lines (SK-BR-3, HBL-100, MDA-MB157, MDA-MB231, and MDA-MB468; Fig. 5).

View larger version (16K): [in this window] [in a new window]   Figure 5 mRNA levels of HRPAP20 and TIMELESS in nine breast tumor cell lines. For each gene, mRNA levels were normalized such that the median value of seven normal human breast tissue samples was 1. The mRNA levels indicated are means from at least three independent experiments.

mRNA expression of HRPAP20 and TIMELESS in normal human tissues

Knowledge of the physiological expression of genes of interest can help to detect abnormal expression in pathological situations. We therefore assessed HRPAP20 and TIMELESS mRNA levels in a collection of 24 normal human tissues (Table 5). The expression of HRPAP20 and TIMELESS was readily detected in all the tissues.

	Discussion

Top Abstract Introduction Materials and methods Results Discussion References

 
In an attempt to identify new candidate molecular markers with which to predict the antiestrogen responsiveness of ER-positive postmenopausal breast cancer, we first used a oligonucleotide microarray approach to measure the expression of a large panel of probes (n=44 290 corresponding to a minimum of 18 795 well-known genes) in ten ER-positive breast tumors from patients who relapsed (n=5) or did not relapse (n=5). Given the relatively small number of the tumor samples, we duplicated the microarray experiments by a dye-swap design. This screening step identified 24 genes that each perfectly discriminated between breast tumors with and without relapse (AUC–ROC, 1.000). These 24 genes of interest were then further investigated in an independent well-defined cohort of 48 ER-positive postmenopausal breast cancer patients treated with primary surgery followed by adjuvant tamoxifen alone; 24 of these patients had relapsed and 24 had not relapsed. Among the 24 genes identified by oligonucleotide microarray analysis, six (HRPAP20, NEK2, TIMELESS, PTPLB, MGC29814, and RANBP2L1) were significantly upregulated in the 24 patients who relapsed as compared with the 24 patients who did not relapse (P<0.05; Table 2). Univariate analysis showed that, among these six genes, only HRPAP20, TIMELESS, PTPLB, and MGC29814 correlated with RFS (Fig. 2A–D).

This two-step strategy was by no means exhaustive, and many possibly relevant genes were certainly missed, but it nevertheless demonstrates that several potentially useful marker genes can be identified in a limited number of microarray experiments.

None of the four genes identified in this study is included in the recently characterized gene expression signatures that predict the clinical outcome of breast cancer patients treated with tamoxifen, i.e., the 44-gene signature by Jansen et al. (2005), the 21-gene signature by Paik et al. (2004) and the two-gene signature by Ma et al. (2004). However, it is difficult to compare our results with those of these three studies, because of differences in the study populations, techniques and materials used. For example, contrary to the authors of these previous studies, we chose to analyze a well-defined cohort of postmenopausal patients with ER-positive breast cancer treated with primary surgery followed by adjuvant tamoxifen alone.

It is noteworthy that we measured the expression of HOXB13 and IL17BR characterized by Ma et al. (2004) in our panel of 48 tumors. In agreement with Reid et al. (2005), we found that HOXB13 and IL17BR expression was similar in the 24 patients who relapsed and the 24 patients who did not relapse: AUC–ROC, 0.505 and 0.482 respectively (data not shown).

Our data identify HRPAP20, TIMELESS, PTPLB, and MGC29814 as candidate prognostic markers in tamoxifen-treated postmenopausal breast cancer patients. However, it remains to be determined whether these four genes predict the tumor response to tamoxifen and/or reflect intrinsic tumor aggressiveness. A prospective randomized study is needed to show whether these molecular markers influence outcome only in patients who receive adjuvant tamoxifen or also in untreated patients. To further investigate the potential endocrine responsiveness function of HRPAP20, TIMELESS, PTPLB, and MGC29814 in ER-positive breast cancer, we analyzed their expression in an OH-Tam-sensitive cell line and in two OH-Tam-resistant cell lines (Badia et al. 2000, Vendrell et al. 2005). We found that two of the four genes (HRPAP20 and TIMELESS) are under direct regulation of E₂ (Figs 3 and 4). Moreover, the E₂ response was abolished by OH-Tam treatment in the OH-Tam-sensitive cell line (antagonistic action of OH-Tam), but not in the two OH-Tam-resistant cell lines (agonistic action of OH-Tam). Taken together, these findings suggest that HRPAP20 and TIMELESS might encode proteins with a critical role in the endocrine responsiveness pathway, but little or nothing is known of the relevance of these two genes in breast cancer biology.

We found only two studies of HRPAP20 from the same team in PubMed journals (Karp et al. 2004, 2007). Interestingly, these authors identified HRPAP20 as a novel phosphoprotein that enhances the growth and survival of hormone responsive tumor cells (Karp et al. 2004). HRPAP20 was identified with a differential mRNA display approach, in a prolactin-dependent rat Nb2 T lymphoma cell line stimulated with various hormones and differentiating agents. Stable transfection of HRPAP20 into MCF-7 cells significantly increased proliferation in the absence of hormone stimulation and augmented survival in the absence of serum (Karp et al. 2004). Very recently, these authors also suggested that HRPAP20 is an important regulator of breast cancer cell invasion by a calmodulin-mediated mechanism leading to increased matrix metalloproteinase 9 secretion (Karp et al. 2007). We also found that HRPAP20 is ubiquitously expressed in human tissues and upregulated in two of the nine breast tumor cell lines tested (MDA-MB361 and MDA-MB468). Finally, in silico analysis of the Oncomine database (http://www.oncomine.org/), that catalogs 300 primary research articles on the cancer transcriptome, showed that HRPAP20 was upregulated in 18 breast cancer 1 (BRCA1) mutation-associated breast tumors as compared with 97 sporadic breast tumors (P=0.0005), based on microarray data from van't Veer et al. (2002).

Little is also known of the role of TIMELESS in cancer biology. Timeless protein is mainly known for its essential role in the circadian rhythm of Drosophila. However, recent human studies suggest an intimate connection between the circadian cycle and DNA damage checkpoints, that is partly mediated by timeless protein (Unsal-Kacmaz et al. 2005). The latter authors also showed that timeless protein interacts with Chk1 kinase, which regulates DNA damage-induced G2/M arrest and is mainly activated by BRCA1 (Yarden et al. 2002). Finally, a recent study (Naderi et al. 2007) identified TIMELESS among a common prognostic signature of 29 genes that associated with survival in breast cancer in three major independent datasets: their cohort (Naderi et al. 2007), the van de Vijver et al. (2002) dataset, and the Wang et al. (2005) dataset.

In the present study, we found that TIMELESS is ubiquitously expressed in human tissues and upregulated in half the breast tumor cell lines tested.

In conclusion, this study points to HRPAP20 and TIMELESS as attractive candidate molecular markers for predicting the tamoxifen responsiveness of ER-positive postmenopausal breast cancer. Further studies are necessary to discover whether these two markers may also be useful in premenopausal breast cancer qualifying for tamoxifen therapy, or for predicting the response to other endocrine agents such as aromatase inhibitors, other selective ER modulators, and progestins. These possibilities are currently being tested in large prospective clinical studies.

	Acknowledgements

 
This work was supported by the Comité départemental des Hauts-de-Seine de la Ligue Nationale Contre le Cancer and the Cancéropole Ile-de-France. We thank the staff of Centre René Huguenin for their assistance in specimen collection and patient care. We are grateful to H RIPOCHE for critical reading of the manuscript. The MVLN, CL6.8, and CL6.32 cells were kindly provided by Dr PONS and Dr BADIA, INSERM U540, Montpellier, France. The authors declare that there is no conflict of interest that would prejudice the impartiality of this scientific work.

	References

Top Abstract Introduction Materials and methods Results Discussion References

 
Badia E, Duchesne MJ, Semlali A, Fuentes M, Giamarchi C, Richard-Foy H, Nicolas JC & Pons M 2000 Long-term hydroxytamoxifen treatment of an MCF-7-derived breast cancer cell line irreversibly inhibits the expression of estrogenic genes through chromatin remodeling. Cancer Research 60 4130–4138.[Abstract/Free Full Text]

Bieche I, Onody P, Laurendeau I, Olivi M, Vidaud D, Lidereau R & Vidaud M 1999 Real-time reverse transcription-PCR assay for future management of ERBB2-based clinical applications. Clinical Chemistry 45 1148–1156.[Abstract/Free Full Text]

Bieche I, Parfait B, Le Doussal V, Olivi M, Rio MC, Lidereau R & Vidaud M 2001 Identification of CGA as a novel estrogen receptor-responsive gene in breast cancer: an outstanding candidate marker to predict the response to endocrine therapy. Cancer Research 61 1652–1658.[Abstract/Free Full Text]

Demirpence E, Duchesne MJ, Badia E, Gagne D & Pons M 1993 MVLN cells: a bioluminescent MCE-7-derived cell line to study the modulation of estrogenic activity. Journal of Steroid Biochemistry and Molecular Biology 46 355–364.[CrossRef][Web of Science][Medline]

DeRisi J, Penland L, Brown PO, Bittner ML, Meltzer PS, Ray M, Chen Y, Su YA & Trent JM 1996 Use of a cDNA microarray to analyse gene expression patterns in human cancer. Nature Genetics 14 457–460.[CrossRef][Web of Science][Medline]

Diab SG, Elledge RM & Clark GM 2000 Tumor characteristics and clinical outcome of elderly women with breast cancer. Journal of the National Cancer Institute 92 550–556.[Abstract/Free Full Text]

Hanley J & McNeil B 1982 The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 143 29–36.[Abstract/Free Full Text]

Jansen MP, Foekens JA, van Staveren IL, Dirkzwager-Kiel MM, Ritstier K, Look MP, Meijer-van Gelder ME, Sieuwerts AM, Portengen H, Dorssers LC et al. 2005 Molecular classification of tamoxifen-resistant breast carcinomas by gene expression profiling. American Journal of Clinical Oncology 23 732–740.[CrossRef]

Kaplan EL & Meier P 1958 Nonparametric estimation from incomplete observations. Journal of the American Statistical Association 53 457–481.[CrossRef][Web of Science]

Karp CM, Pan H, Zhang M, Buckley DJ, Schuler LA & Buckley AR 2004 Identification of HRPAP20: a novel phosphoprotein that enhances growth and survival in hormone-responsive tumor cells. Cancer Research 64 1016–1025.[Abstract/Free Full Text]

Karp CM, Shukla MN, Buckley DJ & Buckley AR 2007 HRPAP20: a novel calmodulin-binding protein that increases breast cancer cell invasion. Oncogene 26 1780–1788.[CrossRef][Web of Science][Medline]

Ma XJ, Wang Z, Ryan PD, Isakoff SJ, Barmettler A, Fuller A, Muir B, Mohapatra G, Salunga R & Tuggle JT 2004 A two-gene expression ratio predicts clinical outcome in breast cancer patients treated with tamoxifen. Cancer Cell 5 607–616.[CrossRef][Web of Science][Medline]

Mann H & Whitney D 1947 On a test of whether one of two random variables is stochastically larger than the other. Annals of Mathematical Statistics 18 50–60.[Web of Science]

McGuire WL 1980 Steroid hormone receptors in breast cancer treatment strategy. Recent Progress in Hormone Research 36 135–156.[Medline]

Naderi A, Teschendorff AE, Barbosa-Morais NL, Pinder SE, Green AR, Powe DG, Robertson JF, Aparicio S, Ellis IO, Brenton JD et al. 2007 A gene-expression signature to predict survival in breast cancer across independent data sets. Oncogene 26 1507–1516.[CrossRef][Web of Science][Medline]

Paik S, Shak S, Tang G, Kim C, Baker J, Cronin M, Baehner FL, Walker MG, Watson D, Park T et al. 2004 A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer. New England Journal of Medicine 351 2817–2826.[Abstract/Free Full Text]

Peto R, Pike MC & Armitage P 1977 Design and analysis of randomized clinical trials requiring prolonged observation of each patient. II. Analysis and examples. British Journal of Cancer 35 1–39.[Web of Science][Medline]

Reid JF, Lusa L, De Cecco L, Coradini D, Veneroni S, Daidone MG, Gariboldi M & Pierotti MA 2005 Limits of predictive models using microarray data for breast cancer clinical treatment outcome. Journal of the National Cancer Institute 97 927–930.[Abstract/Free Full Text]

Unsal-Kacmaz K, Mullen TE, Kaufmann WK & Sancar A 2005 Coupling of human circadian and cell cycles by the timeless protein. Molecular and Cellular Biology 25 3109–3116.[Abstract/Free Full Text]

van't Veer LJ, Dai H, van de Vijver MJ, He YD, Hart AA, Mao M, Peterse HL, van der Kooy K, Marton MJ, Witteveen AT et al. 2002 Gene expression profiling predicts clinical outcome of breast cancer. Nature 415 530–536.[CrossRef][Medline]

Vendrell JA, Bieche I, Desmetz C, Badia E, Tozlu S, Nguyen C, Nicolas JC, Lidereau R & Cohen PA 2005 Molecular changes associated with the agonist activity of hydroxy-tamoxifen and the hyper-response to estradiol in hydroxy-tamoxifen-resistant breast cancer cell lines. Endocrine-Related Cancer 12 75–92.[Abstract/Free Full Text]

van de Vijver MJ, He YD, van't Veer LJ, Dai H, Hart AA, Voskuil DW, Schreiber GJ, Peterse JL, Roberts C, Marton MJ et al. 2002 A gene-expression signature as a predictor of survival in breast cancer. New England Journal of Medicine 347 1999–2009.[Abstract/Free Full Text]

Wang Y, Klijn JG, Zhang Y, Sieuwerts AM, Look MP, Yang F, Talantov D, Timmermans M, Meijer-van Gelder ME, Yu J et al. 2005 Gene-expression profiles to predict distant metastasis of lymph-node-negative primary breast cancer. Lancet 365 671–679.[Web of Science][Medline]

Yarden RI, Pardo-Reoyo S, Sgagias M, Cowan KH & Brody LC 2002 BRCA1 regulates the G2/M checkpoint by activating Chk1 kinase upon DNA damage. Nature Genetics 30 285–289.[CrossRef][Web of Science][Medline]

Received in final form 13 June 2007
Accepted 30 July 2007

This Article Abstract Full Text (PDF) Supplementary data Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Tozlu-Kara, S Articles by Bièche, I Search for Related Content PubMed PubMed Citation Articles by Tozlu-Kara, S Articles by Bièche, I