Clinical Cancer Research CTRC-AACR San Antonio Breast Cancer Symposium
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Meeting Abstracts Online
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplementary Data
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Kreike, B.
Right arrow Articles by van de Vijver, M. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Kreike, B.
Right arrow Articles by van de Vijver, M. J.
Clinical Cancer Research Vol. 12, 5705-5712, October 1, 2006
© 2006 American Association for Cancer Research

Imaging, Diagnosis, Prognosis

Gene Expression Profiles of Primary Breast Carcinomas from Patients at High Risk for Local Recurrence after Breast-Conserving Therapy

Bas Kreike1,3, Hans Halfwerk2,3, Petra Kristel2,3, Annuska Glas2, Hans Peterse2, Harry Bartelink1 and Marc J. van de Vijver2

Authors' Affiliations: Departments of 1 Radiation Oncology, 2 Diagnostic Oncology, and 3 Experimental Therapy, the Netherlands Cancer Institute, Amsterdam, the Netherlands

Requests for reprints: Marc J. van de Vijver, Department of Diagnostic Oncology, the Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX Amsterdam, the Netherlands. Phone: 31-20-512-2750; Fax: 31-20-512-2759; E-mail: m.vd.vijver{at}nki.nl.


   Abstract
Top
 Abstract
Materials and Methods
 Results
 Discussion
 References
 
Purpose: Several risk factors for local recurrence of breast cancer after breast-conserving therapy (BCT) have been identified. The identification of additional risk factors would be very useful in guiding optimal therapy and also in improving understanding of the mechanisms underlying local recurrence. We used cDNA microarray analysis to identify gene expression profiles associated with local recurrence.

Experimental Design: Using 18K cDNA microarrays, gene expression profiles were obtained from 50 patients who underwent BCT. Of these 50 patients, 19 developed a local recurrence; the remaining 31 patients were selected as controls as they were free of local recurrence at least 11 years after treatment. For 9 of 19 patients, the local recurrence was also available for gene expression profiling. Unsupervised and supervised methods of classification were used to separate patients in groups corresponding to disease outcome and to study the overall gene expression pattern of primary tumors and their recurrences.

Results: Hierarchical clustering of patients did not show any grouping reflecting local recurrence status. Supervised analysis revealed no significant set of genes that was able to distinguish recurring tumors from nonrecurring tumors. Paired-data analysis of primary tumors and local recurrences showed a remarkable similarity in gene expression profile between primary tumors and their recurrences.

Conclusions: No significant differences in gene expression between primary breast cancer tumors in patients with or without local recurrence after BCT were identified. Furthermore, analyses of primary tumors and local recurrences show a preservation of the overall gene expression pattern in the local recurrence, even after radiotherapy.

Breast-conserving therapy (BCT) has become the therapy of choice for a large proportion of breast cancer patients. Several randomized controlled trials have shown no difference in survival rates after BCT or mastectomy for stage I and II breast cancer (1–4). Studies comparing the psychological effects of BCT with mastectomy have shown that patients treated with BCT had a better body image, and some studies reported less fear of tumor recurrence in the BCT study group (5, 6). Unfortunately, BCT is associated with a higher rate of local recurrence compared with mastectomy. A local recurrence rate of 10% in 10 years follow-up is generally considered as clinically acceptable for T1-2N0-1 breast cancers. However, local recurrence up to 30% have been reported in young patients (7, 8).

Several risk factors for local recurrence after BCT have been identified: involvement of the surgical margins by invasive carcinoma [relative risk (RR), 3.9-25] and/or extensive ductal carcinoma in situ (RR, 2.5-4.2), young age (RR, 2.8-9.2), and tumor multicentricity (RR, 1.8-3.3) have been found to be associated with high local recurrence rates after BCT (7–16). The addition of a radiation boost after whole breast irradiation and adjuvant systemic therapy reduce local recurrence rates by 40% to 60%, especially in the younger patients (7, 8, 12–15). In addition to the enormous psychological distress for the patient, local recurrence is associated with higher mortality rates due to breast cancer as has been presented by the Early Breast Cancer Trialists' Collaborative Group (17).

Almost all studies report that young age, especially age <50 years, is a major independent risk factor for local recurrence after BCT. Nevertheless, thus far, it is not clear what the underlying (biological) mechanism is for this phenomenon. It would be of great basic and clinical interest to identify additional risk factors for local recurrence in young breast cancer patients.

Gene expression profiling by microarray analysis has shown to be a powerful tool to predict tumor behavior. Studies focusing on several different cancer types have been published showing that gene expression profiling distinguishes groups of patients representing specific tumor types and/or prognosis (18–26). It has been shown that using the gene expression profile of the tumor, prognosis can be more accurately predicted than by clinical variables only (27, 28).

Here, we have used gene expression profiling using microarray analysis to study primary tumors from high-risk patients for local recurrence that underwent BCT. We compared the gene expression pattern of primary tumors that recurred locally to those that did not recur. In addition, we have compared the gene expression pattern of primary tumors that developed a local recurrence to their recurrences.


   Materials and Methods
Top
 Abstract
 Materials and Methods
Results
 Discussion
 References
 
Patient selection and treatment. Breast cancer patients were selected using clinical information collected from the medical registration of the Netherlands Cancer Institute. This is a prospectively collected database containing over 17,000 breast cancer patients who were referred to the departments of radiotherapy and/or surgery of the Netherlands Cancer Institute between January 1980 and June 2005. Patients (n = 8,027) in this database have been treated with BCT and at a median follow-up of 3.9 years (range 0.003-24.03 years); 347 developed a local recurrence. We linked the medical registration database to the fresh-frozen tissue bank database of the Netherlands Cancer Institute and selected every available tumor present in both databases. We then selected tumors from patients who were ages <51 years at the time of treatment, treated by BCT between January 1984 and December 1997, and had no prior malignancies (excluding nonmelanoma skin cancer and dysplasia of the uterine cervix). All tumors were smaller than 5 cm in diameter at pathologic examination (pT1 or pT2); and among those, 19 were from patients who developed a local recurrence and 31 were from patients who continued to be local recurrence-free for ≥11 years after BCT. All patients underwent axillary lymph node dissection, 29 patients had no lymph node metastases, and 21 patients had tumor-involved lymph nodes.

BCT consisted of breast-conserving surgery (segmentectomy or wide local excision) followed by whole breast irradiation to a median dose of 50 Gy (range 46-52 Gy). A radiation boost to the tumor bed was delivered in 45 patients. The boost was applied by means of iridium implantation or external beam irradiation (either photons or electrons) to a median dose of 15 Gy (range 14-25 Gy). Indications for a high boost (>16 Gy) were as follows: focal involvement of tumor in the excision margin or an extensive ductal carcinoma in situ component around the tumor.

This study was approved by the medical ethical committee of the Netherlands Cancer Institute.

Characterization of tumors by histology and immunohistochemistry. For all tumors, histopathologic characteristics were reviewed by one pathologist (M.V.). The features scored included tumor diameter, histologic type, grade, amount of vascular invasion, amount and type of ductal carcinoma in situ component, and resection margin status. Immunohistochemical staining for estrogen receptor (ER), progesterone receptor, HER2, and p53 was done using a tissue microarray for all specimens of which the original paraffin tumor block could be retrieved. Formalin-fixed, paraffin-embedded tissue samples were stained with antibodies against ER (1D5; dilution, 1:150; DAKO, Glostrup, Denmark); progesterone receptor (involving standard antigen retrieval, followed by incubation with progesterone-receptor polyclonal antibody; DAKO), HER2 (3B5; dilution, 1:10,000; ref. 29), and p53 (D07; dilution, 1:8,000; DAKO). Immunohistochemical results were scored semiquantitatively. Tumors were considered positive for hormone receptors if at least 10% of the tumor cells showed nuclear staining. Staining for HER2 was scored as follows: 0, no staining; 1, >10% of cells were weakly positive; 2, moderate homogeneous staining; and 3, strong homogeneous staining.

The tissue microarray was constructed by small core biopsies from selected regions in the paraffin block; three different punched-out cores represented each tumor. All cores were placed in a recipient paraffin block in an ordered fashion and this recipient block was sectioned for immunohistochemical staining. Each tissue element in the tissue microarray is 0.6 mm in diameter and the spacing between two adjacent elements is 1 mm (30).

As not all tumor blocks could be retrieved, and not all tumors were represented in the tissue microarray, we did not obtain immunohistochemistry data for all tumors. Good-quality immunohistochemistry was obtained for 32 primary tumors and 9 recurrences. To compensate for the missing data, we correlated the microarray gene expression data with the available immunohistochemistry for ER and HER2. We then derived a surrogate cutoff level in gene expression for ER and HER2. Samples with an expression ratio for ER <–1.5 were scored as ER negative and samples with an expression ratio for HER2 <0.0 were scored as HER2 negative.

Loss of heterozygosity analysis. We were able to perform loss of heterozygosity analysis to compare the primary tumor and the clinical recurrence for 11 of 19 cases. For these 11 cases, paraffin-embedded tumor blocks of the primary and local recurrence and paraffin-embedded normal tissue of uninvolved lymph nodes were available for DNA isolation. Detailed description of the used loss of heterozygosity technique was previously described by Bosma et al. (31). We tested for loss of heterozygosity on 12 different loci (D13S158, D14S65, D16S3091, D16S520, D17S799, D17S831, D17S849, D17S949, D4S1572, D5S644, D6S446, and D9S288).

Freezing of tumor samples, RNA isolation, and microarray analysis. Tissue samples were snap frozen in liquid nitrogen within 1 hour after surgery. From these frozen tissue blocks, sections were cut for RNA isolation. The first and the last section were used to assess the percentage of tumor cells by H&E staining; only tumors containing on average >40% tumor cells were used in this analysis. Total RNA was isolated with RNAzol B (Campro Scientific, Amersfoort, the Netherlands) and dissolved in RNase-free water. The RNA was treated with DNase. Four micrograms of RNA were amplified and 2 µg aRNA were used for hybridization on the microarray, as described previously by Glas et al. (32).

Glass slide 18K cDNA microarrays were obtained from the central microarray facility at the Netherlands Cancer Institute.4 Tumor samples were cohybridized with reference RNA isolated from a reference pool consisting of over 100 breast cancer samples. Fluorescent intensities were normalized and corrected for a variety of biases that affect the intensity measurements (e.g., color bias and print tip bias) according to Yang et al. (33). Weighted averages and confidence levels were computed according to the Rosetta Error Model (34). Gene expression data has been made publicly available at GEO, series record GSE4913 (http://www.ncbi.nlm.nih.gov/geo).

Data analysis. A subset of the total of 18,432 genes was selected based on the following criteria: expression data should be available for at least 90% of all experiments and their expression level should be significantly different from the reference expression, based on the P value generated by the Rosetta Error Model (34), in at least four experiments with P < 0.01. These criteria reduced the total number of genes from 18,432 to 5,821.

Unsupervised and supervised methods of analysis were done. Average-linkage hierarchical clustering of an uncentered Pearson correlation similarity matrix was applied with the program Cluster and results were visualized with TreeView (35). We have used the previously described "intrinsic" gene set by Sorlie et al. (26) to define basal, luminal A, luminal B, ERBB2, and normal breast-like tumor classes by clustering. For this purpose, we have mapped as many of these intrinsic genes as possible on our microarray-platform (418 of 500 unique UniGene clusters).

We also did hierarchical cluster analysis of the 50 primary tumors with the 5,821-gene subset. In addition, we clustered the 50 primary tumors with the above-described gene subset after omitting estrogen-regulated genes as described by Gruvberger et al. (36), resulting in a list of 5,524 genes.

The supervised analysis was done by using SAM software developed by Tusher et al. (37) We used the settings in the software for two class unpaired data. This approach should result in genes that are differently expressed between the two groups. A threshold was chosen that reflects the lowest false discovery rate (FDR) as estimated after repeatedly permuting (1,000 times) the labels and counting the number of genes that were called significant at each threshold.

We also used SAM to identify the estrogen-regulated genes that were omitted in the hierarchical cluster analysis according to Gruvberger et al. (36).

In addition to the analyses with SAM software, we also used prediction analysis of microarrays developed by Tibshirani et al. (38). This program was developed for class predicting gene signatures.

The paired data set (primary tumor and local recurrence of the same patient) was also analyzed by Cluster, TreeView, and SAM, with settings for paired data.

We have analyzed all tumors from patients that developed local recurrence as a group (n = 19). In addition, we have also analyzed the tumors from patients that developed local recurrence as first event separately (n = 14). This separate analysis was done, because it is known that a subset of patients develops local recurrence concurrently with the development of distant metastases. It may be that tumor-intrinsic risk factors for local recurrence differ for patients that develop an isolated local recurrence. For this separate analysis, patients that developed distant metastases within 3 months after local recurrence were also deemed to have local recurrence simultaneous with distant metastases.


   Results
Top
 Abstract
 Materials and Methods
 Results
Discussion
 References
 
The prospectively collected database of the medical registration of the Netherlands Cancer Institute contains over 17,000 breast cancer patients who were referred to the Departments of Radiotherapy and/or Surgery of the Netherlands Cancer Institute between January 1980 and June 2005. Patients (n = 8,027) in this database have been treated with BCT. In this series of patients, young age was associated with increased risk of local recurrence, in agreement with many previous reports (log-rank P < 0.0001, data not shown). We therefore wanted to identify risk factors for local recurrence especially in young patients and have done gene expression profiling of primary breast cancer tumors from patients ages <51 years at the time of treatment with the aim to identify a gene expression signature associated with local recurrence after BCT. We have selected all tumors from this group of patients when fresh frozen tumor tissue was present, resulting in 19 cases (patients that developed local recurrence) and 31 controls (patients that have remained free of local recurrence for at least 11 years after treatment). In addition, we have analyzed the tumors from only those patients that developed local recurrence as first event (n = 14) to tumors from patients that did not develop a local recurrence (n = 31).

Clinical and pathologic characteristics. Table 1 displays the characteristics of the patients and tumors. To avoid bias, we have not matched the patient groups for known risk factors, because these risk factors may be surrogates for gene expression patterns. From this table, one can clearly see that many clinical and histologic variables are equally distributed over both groups, but that there is a significant larger proportion of younger patients who develop a local recurrence (P = 0.036) and a larger proportion of grade 3 (poorly differentiated) tumors in the group that developed a local recurrence (all local recurrence, P = 0.007; local recurrence as first, P = 0.014). This is as expected, because young age and poor histologic grade are both known risk factors for local recurrence after BCT.


View this table:
[in this window]
[in a new window]
  		Table 1. Patient characteristics

 
Twenty patients received systemic adjuvant therapy; 19 were given six cycles of cyclophosphamide, methotrexate, and 5-fluorouracil chemotherapy. One patient was treated by endocrine therapy only by means of an ovariectomy and two patients received the above-mentioned chemotherapy regimen followed by endocrine therapy. These last three patients all developed a local recurrence.

For 43 patients, the resection margins were found to be free of tumor; focal involvement of the margin by invasive carcinoma was found in four patients (one in the local recurrence group and three in the control group); focal involvement with ductal carcinoma in situ of an extensive in situ component was seen in one patient in the recurrence group and focal involvement of both the in situ and invasive component in two patients (one in each group). More than focal involvement of the invasive or in situ component was not seen in any patient. All seven patients with a positive surgical resection margin received a high radiotherapy boost to compensate for the involved margin.

Table 2 shows that there is an equal distribution of HER2, progesterone receptor, and p53 receptor status between the two groups. However, there is a significantly larger proportion of ER-negative patients that develop a local recurrence compared with the control group (all local recurrence, P = 0.007; local recurrence as first, P = 0.037). The receptor status based on gene expression level (compensating for missing data in the tissue microarray data) also shows a significant larger proportion of ER-negative tumors present in the group of patients developing a local recurrence (all local recurrence, P = 0.001; local recurrence as first, P = 0.003).


View this table:
[in this window]
[in a new window]
  		Table 2. Immunohistochemistry of 50 primary tumors and ER and HER2 receptor status based on gene expression data as measured on the microarray for 50 primary tumors (cutoff, ER < –1.5; HER2 < 0.0)

 
When a recurrent tumor develops after BCT, there always is the possibility of a second primary carcinoma. Based on available data about the localization of the recurrence with respect to the primary tumor combined with the histologic and immunohistochemical features, all 19 recurrent tumors had the characteristics of a local recurrence rather than a second primary tumor. In addition, we did loss of heterozygosity analysis on the primary tumor and the local recurrence using markers for 12 unique loci and confirmed the likelihood of a true recurrence instead of a second primary, whereas the recurrence showed loss of one or several particular loci identical to the primary tumor. In some cases, the recurrence showed additional loss of loci compared with the primary tumor.

Hierarchical clustering based on gene expression in the primary tumors. We did hierarchical cluster analysis of 50 primary breast carcinomas using the intrinsic gene set described by Sorlie et al. (26). This analysis showed grouping of the earlier published molecular subtypes. We calculated the Pearson correlation of the 50 primary tumors to the centroids for each of the five molecular subtypes. We assigned each tumor to the subtype with the highest correlation. Forty tumors could be assigned to one of the five molecular subtypes based on the correlation of its average gene expression to the centroid of the subtype. However, the clustering showed no grouping of the local recurrence status (P = 0.29); there was also no significant distribution of local recurrence status among the 40 assigned subtypes (P = 0.74; Fig. 1 ).


Figure 1
View larger version (82K):
[in this window]
[in a new window]
  		Fig. 1. Two-dimensional hierarchical clustering of 50 primary tumors and 418 genes of the intrinsic gene list. A, heat map of 50 primary tumors at the top and 418 genes at the side. Branches are color-coded according to the subtypes to which the corresponding tumor shows the highest correlation: basal (red), luminal A (dark blue), luminal B (light blue), ERRB2 (pink), and normal breast-like (green); tumors with low correlation (<0.1) to any subtype (gray branches). B, zoomed-in image of the patient's dendogram (rotated 90° clockwise) with columns at the left-hand side indicating, respectively, local recurrence status (1 = yes, 0 = no), age (years), histologic grade, and ER status; groups are highlighted by the colored boxes.

 
Two-dimensional hierarchical cluster analysis of the 50 primary tumors with the 5,821 genes subset also showed no grouping of the local recurrence status (data not shown). The two main clusters seem to be largely based on ER status ({chi}2 test, P = 0.002), as is usually found in series of breast carcinomas, whereas the P value for local recurrence status is 0.12. Gruvberger et al. (36) have shown that there is a set of genes responsible for the ER cluster dominancy. We have identified this set of genes in our data set by performing a two-class SAM analysis of unpaired data on the ER status of the tumors. A set of 297 genes was significantly (median FDR = 0.21) regulated with ER status. When correcting for the ER-regulated genes by removing them from the list of 5,821 genes, we repeated the clustering of 50 tumors and 5,524 genes. Using this approach, the clustering was not ER driven, reflected by a nonsignificant distribution of ER status in both main branches of the cluster dendogram (P = 0.70); there was also no significant grouping of the local recurrence status, P = 0.96 (data not shown).

Supervised analysis based on the gene expression in the primary tumors. We have used a supervised method to distinguish genes differently expressed between the two study groups, i.e., tumors with and without a local recurrence during follow-up. The two-class SAM analysis for unpaired data resulted in a lowest median FDR of 3.56%, resulting in a gene list of 23 genes (Supplementary Data). The top-scoring gene in this list is fatty acid binding protein 3 (FABP3). The third best-scoring gene is ER-{alpha}. As ER status has been found to be associated with primary tumors developing a local recurrence, this gene expression pattern in tumors from patients who developed a local recurrence may be biased by ER status. We therefore did an additional SAM analysis on only the ER-positive patients (n = 39). This resulted in a lowest median FDR of 19.10%, for a gene list of 106 genes (Supplementary Data). In view of the high FDR, it has to be concluded that there is no significant difference in gene expression pattern between the two groups of tumors.

In addition, we also did a two-class SAM analysis on the 50 primary tumors and the reduced gene set after filtering out the ER-regulated genes (n = 297). This resulted in a list of 28 genes with a lowest median FDR of 9.14%, again too high for significance (Supplementary Data).

Class prediction was done by using prediction analysis of microarrays. Initially, we used all 50 primary tumors and 5,821 genes for class prediction. The cross-validation procedure showed an optimal performance with 94 genes with a misclassification error of 42% for tumors that recurred locally and 29% for tumors that remained free of recurrence. To evaluate the performance of the classification procedure, we divided our group into a training and test group at random in a 2/3 to 1/3 ratio. The training cross-validation procedure showed an optimal performance with 109 genes with a misclassification error of 36% for local recurrence tumors and 24% for tumors that remained free of recurrence. The validation of this 109-gene local recurrence-predictor showed 100% correct prediction of no-local recurrence tumors, but only 12.5% correct prediction of local recurrence tumors (Supplementary Data).

Supervised analysis for 45 primary tumors of patients who developed local recurrence as first event. As the biology of local recurrence as first event may be different from local recurrence in conjunction with distant metastases, we tested our data set for a difference in gene expression between tumors that develop a local recurrence as first event versus tumors that remain free from local recurrences in the same way as we have described above. It is the same set of tumors, but now we have excluded those cases where the local recurrence did not occur at least 3 months before any other event.

The two class SAM analysis on 45 primary tumors resulted in a lowest median FDR of 14.29% with a list of seven genes (Supplementary Data). Again, in view of the high FDR, there does not seem to be a reliable gene expression signature for local recurrence in this series of patients.

Gene expression in nine pairs of primary tumors and local recurrences. In addition to the analysis of the primary tumors, we were able to retrieve the actual fresh-frozen recurrence in 9 of the possible 19 cases. Using unsupervised hierarchical cluster analysis, we compared the overall gene expression pattern of these nine primary tumors and nine patient-matched local recurrences. The hierarchical cluster analysis showed close clustering of recurrences with their primary tumors in almost all cases (Fig. 2 ). This means that the overall gene expression of a local recurrence is more similar to its primary tumor than to other primary tumors or local recurrences. This is interesting because all local recurrences, compared with all primary tumors, have been irradiated up to a dose of at least 50 Gy.


Figure 2
View larger version (32K):
[in this window]
[in a new window]
  		Fig. 2. Two-dimensional hierarchical clustering of nine pairs (primary tumor and local recurrence). A, heat map of nine primary tumors and nine recurrences at the top and 5,821 genes at the side. B, zoomed-in image of the tumor dendogram (rotated 90° clockwise) with corresponding labels; the clustering of a pair is highlighted by the red box around the tumor pair. LR, local recurrence; prim, primary tumor.

 
The result of the hierarchical cluster analysis is confirmed by a supervised SAM analysis for paired data. In this type of analysis, differences in gene expression between primary tumors and their local recurrences can be investigated. The analysis resulted in a lowest median FDR of 29.85% (Supplementary Data). This indicates that there is no set of genes that is consistently different in expression between primary tumors and recurrences. The pairs are more related to each other than the primary tumors or recurrences as a group are.


   Discussion
Top
 Abstract
 Materials and Methods
 Results
 Discussion
References
 
We have used gene expression profiling to study invasive breast carcinomas from patients <51 years who underwent BCT. The aim of the study was to identify a gene expression signature, associated with local recurrence after BCT.

When analyzing the primary tumors only, we found no substantial separation of primary tumors from local recurrence-free patients from tumors that recurred in the treated breast. We realize that the number of tumors for this study was relatively small, but for this analysis we used all available fresh frozen tumor material present at the Netherlands Cancer Institute of primary tumors treated with BCT and follow-up data of at least 11 years. If there would have been a strong local recurrence predictive profile, however, it should have been detected by our approach.

The hierarchical cluster analysis of the 50 primary tumors with the intrinsic gene set did show clusters of specific breast cancer subtypes, as reported previously by Perou et al. (23) and Sorlie et al. (26). We did not observe significant grouping of the tumors that were similar with respect to their local recurrence status. Clustering with a larger gene set also did not reveal any grouping, nor after adjusting the gene set for estrogen-regulated genes. We think that this indicates that there is not a large difference in overall gene expression that accounts for local recurrence in breast cancer.

In the supervised classification analysis of 50 primary tumors, we have found a set of genes that was significantly different in gene expression with reasonably low FDRs. We think that this result is biased by the significant difference in distribution of ER-negative tumors, because one of the top-scoring genes in the analysis was the ER itself. We have tried to overcome this problem by adjusting the gene list for estrogen-regulated genes according to the approach previously described by Gruvberger et al. (36) and in a second analysis by only focusing the analysis on ER-positive tumors. However, both methods resulted without obtaining a reliable local recurrence classifier.

We also tried to construct a class-predicting gene expression signature by using the principle of nearest shrunken centroids to identify subsets of genes that best characterize each class (prediction analysis of microarrays). Although the initial trained classifier resulted in a reasonable performance, the validation of this classifier showed poor performance. This phenomenon of reasonable training performance but poor validation performance is often seen in constructing classifiers using a relatively small number of samples. To overcome this problem, a larger sample size will be required in future studies.

Although the SAM analyses all resulted in high FDRs, the gene expression pattern that was most associated with local recurrence was FABP3. The expression level of this gene was significantly more up-regulated in tumors from patients that recurred in the breast within 11 years after therapy. FABP3 is also known as mammary-derived growth inhibitor and is a candidate tumor suppressor gene for human breast cancer. In vitro studies have shown that FABP3 inhibits growth of normal mammary epithelial cells (39, 40) and that it is highly expressed in terminally differentiated mammary cells in contrast to proliferating tissue from pregnant animals (41, 42). However, FABP3 fails to inhibit proliferation of some breast cancer cell lines (43, 44).

In 2000, Voehringer et al. (45) showed that a family of FABPs play a role in the control of resistance to apoptosis after exposure to radiation in a B-cell lymphoma model system. They showed increased gene expression of a family of FABP genes in the cells that were radioresistant. They postulated that FABPs bind the fatty acid retinoic acid and thereby create apoptosis-resistant cells. They further hypothesized that FABPs may act as molecular scavengers, binding toxic-oxidized fatty acids generated during oxidative stress, thus terminating a cascade of lipid peroxidation, observed as an early step in the commitment phase of many apoptotic systems. In our data, the mean expression level of FABP3 is higher in the group of patients that developed a local recurrence. As increased levels of FABP3 could be associated with resistance to radiotherapy, there may be a causative role for FABP3 in local recurrence after BCT.

We also compared the gene expression pattern of primary tumors with their local recurrences. The paired data showed that in the majority of cases, the recurrences cluster nicely next to their primary tumor. The supervised analysis comparing primary tumors and local recurrences resulted in high FDRs, indicating that there is no significant difference in gene expression pattern between the primary tumors and local recurrences. This is remarkable, although treatment of the primary tumor site always included radiotherapy up to a dose of at least 50 Gy. From our data, it can be concluded that the overall gene expression profile in the tumor is not affected by undergoing the exposure of irradiation.

Breast cancer is a heterogeneous disease and the identification of diagnostic tools to tailor treatment for individual patients is an important goal. Gene expression profiling is a very promising approach to identify such diagnostic tools and has been shown to be successful for the identification of prognostic gene expression profiles (18, 27, 28). Prediction of response to systemic treatment of breast cancer using gene expression profiling has also been attempted in a number of relatively small studies: Ayers et al. (46) and Chang et al. (47) identified classifiers associated with response to neoadjuvant chemotherapy, but Hannemann et al. (48) could not identify such a profile; and profiles predicting recurrence after adjuvant tamoxifen therapy have been identified by Jansen et al. (49) and Ma et al. (50).

The goal is to develop and combine gene expression profiles that can predict risk of distant metastases, risk of local recurrence, and profiles that are associated with response to specific systemic treatment regimens.

Although we were unable to find a strong classifier for local recurrence after BCT in this data set, it cannot be concluded that such a classifier for the prediction of local recurrence does not exist. The main conclusion is that the differences in gene expression between tumors of patients that developed a local recurrence and those that remain free of local recurrence are not sufficiently large to be detected in a relatively small study.

In conclusion, we found no great differences in gene expression between primary breast cancer tumors in young women with or without local recurrence after BCT. Furthermore, analyses of primary tumors and local recurrences in a paired model show a preservation of the overall gene expression pattern in the recurred tumor, even after radiotherapy.


   Acknowledgments
 
We thank Tony van de Velde for the clinical data; Ron Kerkhoven, Arno Velds, Mike Heimerikx, and Daoud Sie from the microarray facility at the Netherlands Cancer Institute; and Guus Hart, Tibor van Welsum, Cathy Bosch, Juliane Hannemann, Anke Witteveen, Leonie Delahaye, and Dimitry Nuyten for helpful suggestions.


   Footnotes
 
Grant support: The Dutch Cancer Society.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).

4 Detailed information on these arrays can be found at http://microarrays.nki.nl/. Back

Received 4/ 3/06; revised 5/23/06; accepted 6/22/06.


   References
Top
 Abstract
 Materials and Methods
 Results
 Discussion
 References
 

  1. Blichert-Toft M, Rose C, Andersen JA, et al. Danish randomized trial comparing breast conservation therapy with mastectomy: six years of life-table analysis. Danish Breast Cancer Cooperative Group. J Natl Cancer Inst Monogr 1992;11:19–25.[Medline]
  2. Fisher B, Anderson S, Bryant J, et al. Twenty-year follow-up of a randomized trial comparing total mastectomy, lumpectomy, and lumpectomy plus irradiation for the treatment of invasive breast cancer. N Engl J Med 2002;347:1233–41.[Abstract/Free Full Text]
  3. van Dongen JA, Voogd AC, Fentiman IS, et al. Long-term results of a randomized trial comparing breast-conserving therapy with mastectomy: European Organization for Research and Treatment of Cancer 10801 trial. J Natl Cancer Inst 2000;92:1143–50.[Abstract/Free Full Text]
  4. Veronesi U, Salvadori B, Luini A, et al. Breast conservation is a safe method in patients with small cancer of the breast. Long-term results of three randomised trials on 1,973 patients. Eur J Cancer 1995;31A:1574–9.[CrossRef]
  5. Bartelink H, van Dam F, van Dongen J. Psychological effects of breast conserving therapy in comparison with radical mastectomy. Int J Radiat Oncol Biol Phys 1985;11:381–5.[Medline]
  6. Curran D, van Dongen JP, Aaronson NK, et al. Quality of life of early-stage breast cancer patients treated with radical mastectomy or breast-conserving procedures: results of EORTC Trial 10801. The European Organization for Research and Treatment of Cancer (EORTC), Breast Cancer Co-operative Group (BCCG). Eur J Cancer 1998;34:307–14.[CrossRef][Medline]
  7. Bartelink H, Horiot JC, Poortmans P, et al. Recurrence rates after treatment of breast cancer with standard radiotherapy with or without additional radiation. N Engl J Med 2001;345:1378–87.[Abstract/Free Full Text]
  8. Park CC, Mitsumori M, Nixon A, et al. Outcome at 8 years after breast-conserving surgery and radiation therapy for invasive breast cancer: influence of margin status and systemic therapy on local recurrence. J Clin Oncol 2000;18:1668–75.[Abstract/Free Full Text]
  9. Borger J, Kemperman H, Hart A, Peterse H, van Dongen J, Bartelink H. Risk factors in breast-conservation therapy. J Clin Oncol 1994;12:653–60.[Abstract]
  10. Elkhuizen PH, van de Vijver MJ, Hermans J, Zonderland HM, van de Velde CJ, Leer JW. Local recurrence after breast-conserving therapy for invasive breast cancer: high incidence in young patients and association with poor survival. Int J Radiat Oncol Biol Phys 1998;40:859–67.[CrossRef][Medline]
  11. Elkhuizen PH, Voogd AC, van den Broek LC, et al. Risk factors for local recurrence after breast-conserving therapy for invasive carcinomas: a case-control study of histological factors and alterations in oncogene expression. Int J Radiat Oncol Biol Phys 1999;45:73–83.[CrossRef][Medline]
  12. Elkhuizen PH, van Slooten HJ, Clahsen PC, et al. High local recurrence risk after breast-conserving therapy in node- negative premenopausal breast cancer patients is greatly reduced by one course of perioperative chemotherapy: A European Organization for Research and Treatment of Cancer Breast Cancer Cooperative Group Study. J Clin Oncol 2000;18:1075–83.[Abstract/Free Full Text]
  13. Voogd AC, Peterse JL, Crommelin MA, et al. Histological determinants for different types of local recurrence after breast-conserving therapy of invasive breast cancer. Dutch Study Group on local Recurrence after Breast Conservation (BORST). Eur J Cancer 1999;35:1828–37.[CrossRef][Medline]
  14. Vrieling C, Collette L, Fourquet A, et al. Can patient-, treatment- and pathology-related characteristics explain the high local recurrence rate following breast-conserving therapy in young patients? Eur J Cancer 2003;39:932–44.[CrossRef][Medline]
  15. Fredriksson I, Liljegren G, Palm-Sjovall M, et al. Risk factors for local recurrence after breast-conserving surgery. Br J Surg 2003;90:1093–102.[CrossRef][Medline]
  16. Voogd AC, Nielsen M, Peterse JL, et al. Differences in risk factors for local and distant recurrence after breast-conserving therapy or mastectomy for stage I and II breast cancer: pooled results of two large European randomized trials. J Clin Oncol 2001;19:1688–97.[Abstract/Free Full Text]
  17. Clarke M, Collins R, Darby S, et al. Effects of radiotherapy and of differences in the extent of surgery for early breast cancer on local recurrence and 15-year survival: an overview of the randomised trials. Lancet 2005;366:2087–106.[Medline]
  18. 't Veer LJ, Dai H, van de Vijver MJ, et al. Gene expression profiling predicts clinical outcome of breast cancer. Nature 2002;415:530–6.[CrossRef][Medline]
  19. Ahr A, Karn T, Solbach C, et al. Identification of high risk breast-cancer patients by gene expression profiling. Lancet 2002;359:131–2.[CrossRef][Medline]
  20. Beer DG, Kardia SL, Huang CC, et al. Gene-expression profiles predict survival of patients with lung adenocarcinoma. Nat Med 2002;8:816–24.[Medline]
  21. Bittner M, Meltzer P, Chen Y, et al. Molecular classification of cutaneous malignant melanoma by gene expression profiling. Nature 2000;406:536–40.[CrossRef][Medline]
  22. Kihara C, Tsunoda T, Tanaka T, et al. Prediction of sensitivity of esophageal tumors to adjuvant chemotherapy by cDNA microarray analysis of gene-expression profiles. Cancer Res 2001;61:6474–9.[Abstract/Free Full Text]
  23. Perou CM, Jeffrey SS, van de Rijn M, et al. Distinctive gene expression patterns in human mammary epithelial cells and breast cancers. Proc Natl Acad Sci U S A 1999;96:9212–7.[Abstract/Free Full Text]
  24. Perou CM, Sorlie T, Eisen MB, et al. Molecular portraits of human breast tumours. Nature 2000;406:747–52.[CrossRef][Medline]
  25. Ramaswamy S, Ross KN, Lander ES, Golub TR. A molecular signature of metastasis in primary solid tumors. Nat Genet 2003;33:49–54.[CrossRef][Medline]
  26. Sorlie T, Perou CM, Tibshirani R, et al. Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc Natl Acad Sci U S A 2001;98:10869–74.[Abstract/Free Full Text]
  27. van de Vijver MJ, He YD, van't Veer LJ, et al. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med 2002;347:1999–2009.[Abstract/Free Full Text]
  28. Wang Y, Klijn JG, Zhang Y, et al. Gene-expression profiles to predict distant metastasis of lymph-node-negative primary breast cancer. Lancet 2005;365:671–9.[Medline]
  29. van de Vijver MJ, Peterse JL, Mooi WJ, et al. Neu-protein overexpression in breast cancer. Association with comedo-type ductal carcinoma in situ and limited prognostic value in stage II breast cancer. N Engl J Med 1988;319:1239–45.[Abstract]
  30. Kononen J, Bubendorf L, Kallioniemi A, et al. Tissue microarrays for high-throughput molecular profiling of tumor specimens. Nat Med 1998;4:844–7.[CrossRef][Medline]
  31. Bosma AJ, Weigelt B, Lambrechts AC, et al. Detection of circulating breast tumor cells by differential expression of marker genes. Clin Cancer Res 2002;8:1871–7.[Abstract/Free Full Text]
  32. Glas AM, Kersten MJ, Delahaye LJ, et al. Gene expression profiling in follicular lymphoma to assess clinical aggressiveness and to guide the choice of treatment. Blood 2005;105:301–7.[Abstract/Free Full Text]
  33. Yang YH, Dudoit S, Luu P, et al. Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation. Nucleic Acids Res 2002;30:e15.[Abstract/Free Full Text]
  34. Hughes TR, Marton MJ, Jones AR, et al. Functional discovery via a compendium of expression profiles. Cell 2000;102:109–26.[CrossRef][Medline]
  35. Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci U S A 1998;95:14863–8.[Abstract/Free Full Text]
  36. Gruvberger S, Ringner M, Chen Y, et al. Estrogen receptor status in breast cancer is associated with remarkably distinct gene expression patterns. Cancer Res 2001;61:5979–84.[Abstract/Free Full Text]
  37. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A 2001;98:5116–21.[Abstract/Free Full Text]
  38. Tibshirani R, Hastie T, Narasimhan B, Chu G. Diagnosis of multiple cancer types by shrunken centroids of gene expression. Proc Natl Acad Sci U S A 2002;99:6567–72.[Abstract/Free Full Text]
  39. Kurtz A, Spitzer E, Zschiesche W, Wellstein A, Grosse R. Local control of mammary gland differentiation: mammary-derived growth inhibitor and pleiotrophin. Biochem Soc Symp 1998;63:51–69.[Medline]
  40. Yang Y, Spitzer E, Kenney N, et al. Members of the fatty acid binding protein family are differentiation factors for the mammary gland. J Cell Biol 1994;127:1097–109.[Abstract/Free Full Text]
  41. Binas B, Gusterson B, Wallace R, Clark AJ. Epithelial proliferation and differentiation in the mammary gland do not correlate with cFABP gene expression during early pregnancy. Dev Genet 1995;17:167–75.[CrossRef][Medline]
  42. Politis I, Gorewit RC, Muller T, Grosse R. Mammary-derived growth inhibitor in lactation and involution. Domest Anim Endocrinol 1992;9:89–94.[CrossRef][Medline]
  43. Lehmann W, Widmaier R, Langen P. Response of different mammary epithelial cell lines to a mammary derived growth inhibitor (MDGI). Biomed Biochim Acta 1989;48:143–51.[Medline]
  44. Wang HL, Kurtz A. Breast cancer growth inhibition by delivery of the MDGI-derived peptide P108. Oncogene 2000;19:2455–60.[CrossRef][Medline]
  45. Voehringer DW, Hirschberg DL, Xiao J, et al. Gene microarray identification of redox and mitochondrial elements that control resistance or sensitivity to apoptosis. Proc Natl Acad Sci U S A 2000;97:2680–5.[Abstract/Free Full Text]
  46. Ayers M, Symmans WF, Stec J, et al. Gene expression profiles predict complete pathologic response to neoadjuvant paclitaxel and fluorouracil, doxorubicin, and cyclophosphamide chemotherapy in breast cancer. J Clin Oncol 2004;22:2284–93.[Abstract/Free Full Text]
  47. Chang JC, Wooten EC, Tsimelzon A, et al. Gene expression profiling for the prediction of therapeutic response to docetaxel in patients with breast cancer. Lancet 2003;362:362–9.[CrossRef][Medline]
  48. Hannemann J, Oosterkamp HM, Bosch CA, et al. Changes in gene expression associated with response to neoadjuvant chemotherapy in breast cancer. J Clin Oncol 2005;23:3331–42.[Abstract/Free Full Text]
  49. Jansen MP, Foekens JA, van Staveren IL, et al. Molecular classification of tamoxifen-resistant breast carcinomas by gene expression profiling. J Clin Oncol 2005;23:732–40.[Abstract/Free Full Text]
  50. Ma XJ, Wang Z, Ryan PD, et al. A two-gene expression ratio predicts clinical outcome in breast cancer patients treated with tamoxifen. Cancer Cell 2004;5:607–16.[CrossRef][Medline]



This article has been cited by other articles:


Home page
 BloodHome page
D. Cervi, T.-T. Yip, N. Bhattacharya, V. N. Podust, J. Peterson, A. Abou-Slaybi, G. N. Naumov, E. Bender, N. Almog, J. E. Italiano Jr, et al.
Platelet-associated PF-4 as a biomarker of early tumor growth
Blood, February 1, 2008; 111(3): 1201 - 1207.
[Abstract] [Full Text] [PDF]

Home page
 JNCI J Natl Cancer InstHome page
M. A. Bollet, N. Servant, P. Neuvial, C. Decraene, I. Lebigot, J.-P. Meyniel, Y. De Rycke, A. Savignoni, G. Rigaill, P. Hupe, et al.
High-Resolution Mapping of DNA Breakpoints to Define True Recurrences Among Ipsilateral Breast Cancers
J Natl Cancer Inst, January 2, 2008; 100(1): 48 - 58.
[Abstract] [Full Text] [PDF]

Home page
 Clin. Cancer Res.Home page
P. Smith, L. J. Nicholson, N. Syed, A. Payne, L. Hiller, O. Garrone, M. Occelli, M. Gasco, and T. Crook
Epigenetic Inactivation Implies Independent Functions for Insulin-like Growth Factor Binding Protein (IGFBP)-Related Protein 1 and the Related IGFBPL1 in Inhibiting Breast Cancer Phenotypes
Clin. Cancer Res., July 15, 2007; 13(14): 4061 - 4068.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplementary Data
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Kreike, B.
Right arrow Articles by van de Vijver, M. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Kreike, B.
Right arrow Articles by van de Vijver, M. J.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Meeting Abstracts Online