|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Originally published as JCO Early Release 10.1200/JCO.2005.03.8802 on April 24 2006 © 2006 American Society of Clinical Oncology. Genes Associated With Breast Cancer Metastatic to Bone
From the Department of Medical Oncology, Erasmus MC–Daniel den Hoed, Rotterdam, the Netherlands; and Veridex LLC, a Johnson & Johnson Company, San Diego, CA Address reprint requests to John A. Foekens, PhD, Erasmus MC, Josephine Nefkens Institute, Rm BE-426, Dr Molewaterplein 50, 3015 GE Rotterdam, the Netherlands; e-mail: j.foekens{at}erasmusmc.nl
Purpose The biology of tumors relapsing to bone is poorly understood. In this study, we initiated a search for genes that are implicated in tumors relapsing to bone in breast cancer. Patients and Methods We analyzed 107 primary breast tumors in patients who were all lymph node negative at the time of diagnosis and all had experienced relapse. Total RNA isolated from frozen tumor samples was used to gather gene expression data using oligo microarrays. Results A panel of 69 genes was found significantly differentially expressed between patients who experienced relapse to bone versus those who experienced relapse elsewhere in the body. The most differentially expressed gene, TFF1, was confirmed by quantitative reverse transcriptase polymerase chain reaction in an independent cohort (n = 122; P = .0015). Our differentially expressed genes, combined with a recently reported gene set relevant to tumors relapsing to bone in an animal model system, pointed to the involvement of the fibroblast growth factor receptor signaling pathway in preference of tumor cells that relapse to bone. Given that patients who experience relapse to bone may benefit from bisphosphonate therapy, we developed a classifier of 31 genes, which in an independent validation set correctly predicts all tumors relapsing to bone with a specificity of 50%. Conclusion Our study identifies a panel of genes relevant to bone metastasis in breast cancer. The subsequently developed classifier of tumors relapsing to bone could, after thorough confirmation on an extended number of independent samples, and in combination with our previously developed high-risk profile, provide a diagnostic tool for the recommendation of adjuvant bisphosphonate therapy in addition to endocrine therapy or chemotherapy.
The most abundant site of a distant relapse in breast, prostate, thyroid, kidney, and lung cancer patients is the bone. Breast cancer mainly gives rise to osteolytic bone lesions and it is plausible that the bone microenvironment facilitates circulating cancer cells to home and proliferate. Many factors have been implicated in facilitating relapse of tumors to bone, including blood flow in red bone marrow, immobilized growth factors (transforming growth factors β, bone morphogenetic proteins, platelet-derived growth factor, insulin-like growth factors, and fibroblast growth factors [FGFs]) in the bone matrix, and adhesive molecules on the tumor cells.1-4 However, in breast cancer patients the genes in the cancer cells promoting the interactions with bone are largely unknown. From the analyzed biologic factors detectable in primary tumors, it is the estrogen receptor (ER) status that currently has the strongest association with metastasis to the bone. We recently described a 76-gene signature able to identify lymph node–negative breast cancer patients at high risk of distant recurrence.5 The ability of a tumor to metastasize, however, is a different event than the tumor’s ability to home, adhere, extravasate, and proliferate in a certain organ. This is concluded from the fact that the site of relapse is not associated with metastasis-free survival time or with our 76-gene prognostic signature5 (unpublished observation). These observations in patients are further supported by in vivo experiments using human breast cancer xenographs in nude mice.6,7 Therefore, using microarray gene expression data (Affymetrix U133A, Santa Clara, CA), we examined 107 samples from which the site of relapse was known, assuming that differentially expressed genes between tumors relapsing to bone or elsewhere will give us insight into the biologic features present in the primary breast cancer cells to preferentially interact with the bone microenvironment. Finally, we extracted a predictor of tumors relapsing to bone that may have a clinical application. Based on the fact that the identified predictor assigns patients who will develop a relapse into a group that experiences relapse to bone or elsewhere, this bone profile is only applicable after patients have been identified as being at high risk for a relapse (eg, using our 76-gene profile). When the bone profile is properly validated in independent cohorts, it may be used to include bisphosphonates as an additional adjuvant systemic therapy in early breast cancer to prevent bone metastasis.8
Patient Data and Gene Expression Analysis The study was performed in accordance to the Code of Conduct of the Federation of Medical Scientific Societies in the Netherlands (www.fmwv.nl). Relevant patient and tumor characteristics are presented in Table 1. The gene expression data (Affymetrix U133a GeneChip) used in this study were taken from our earlier study,5 using 17,819 probe sets that were designated as being present in two or more samples. For each probe set, the signal was expressed relative to the geometric mean of that probe set and was 2-log transformed. We included samples from those patients from which the site of relapse was known (n = 107). Samples were classified as bone when patients developed a tumor relapsing to bone, which included those who had additional metastases in other parts of the body. The remaining samples were labeled nonbone.
Data Analysis Significance analysis of microarrays (SAM) analysis was used to identify differentially expressed genes.9 Three hundred permutations of the data were used to calculate a false discovery rate. Genes were considered significant when the false discovery rate was less than 5% and when a minimum of 1.7-fold difference in expression level was observed. To construct a diagnostic profile, samples were divided into a training set (n = 72; 46 with a relapse to bone and 26 with a relapse in a nonbone area) and a testing set (n = 35; 23 with a relapse to bone and 12 with a relapse to a nonbone area) stratified by site of relapse, ER protein level, and metastasis-free interval. A gene selection step using an optimal cutoff procedure was performed in the samples of the training set. All measured expression levels of a gene were used as the cut point to assign the gene expression as being high or low in a particular sample; a minimum of 20 samples were kept in one of the groups.
Because we knew the site of relapse of these samples, we counted the frequencies for the categories high bone, low bone, high nonbone, and low nonbone for each cutoff. The optimal cutoff was determined by using the TFF1 protein levels were assessed previously 12 in cytosols from tumor tissues and were measured as nanograms TFF1 per milligram of protein; these data were analyzed in a like manner as for the RT-PCR data. Unlike the patients from the microarray study, this cohort also contained node-positive patients but none received adjuvant hormonal or chemotherapy. Statistical analyses were performed using Analyze-it software (Analyze-it Software Ltd, Leeds, United Kingdom).
Identification of Differentially Expressed Genes We used previously retrieved gene expression data5 from 107 lymph node–negative primary breast tumors of patients who had developed distant metastasis. Samples were classified according to the site of relapse: 69 samples were labeled as bone and 38 samples were labeled as nonbone. Using SAM,9 we considered 73 probe-sets representing 69 unique genes as significantly differentially expressed between the bone and nonbone samples. The five highest ranking genes were TFF1, TFF3, AGR2, NAT1, and CRIP1, all of which are more highly expressed in the samples from patients with relapse to bone (Table 2). The highest ranked gene, TFF1, was studied in 122 independent breast tumors from node-negative patients with comparable clinical characteristics by quantitative RT-PCR. TFF1 expression was associated significantly with the site of relapse (P = .0015), with relative median expression level for TFF1 2-log scale of 3.02 (95% CI, 1.41 to 4.66) and –1.63 (95% CI, –5.44 to 2.49) for the group with relapse to bone and nonbone areas, respectively. In addition, the protein levels measured previously in 610 tumors from node-negative and node-positive patients12 showed a higher TFF1 protein content in the patients who experienced relapse to bone (P = .012). Median level and 95% CI was 46.18 (37.33 to 55.03) for the group with relapse to bone and 38.33 (27.11 to 49.56) for the group with relapse to a nonbone area.
Development of a Predictor of Tumor Relapse to Bone Given that the results indicate that the gene expression profile from primary tumors relapsing to bone is considerably different from that of tumors relapsing elsewhere in the body, we explored whether it is possible to predict tumor relapse to bone using these gene expression data. To study this, the samples were divided into a training set (n = 72) and a testing set (n = 35) stratified by site of relapse, ER protein level, and metastasis-free interval. Using the optimal cutoff procedure, we selected 588 informative genes for input in a PAM analysis.10 A 31-gene predictor was selected after 10-fold cross validation in the training set that could identify the samples with tumor relapse to bone in the testing set with 100% sensitivity and 50% specificity. The predictor showed a 79.3% positive predictive value and misclassified 17.1% of the samples. 17 genes in the profile, including TFF1, were also present in the SAM gene list (all 31 genes are referenced in the PAM column in Table 2). To ascertain the validity of our gene set, we analyzed 50 sets of 100 randomly chosen genes. These random gene sets were used for input in a PAM analysis using the same training and testing set. Twenty-nine of the 50 random gene sets showed 100% sensitivity. Of these 29 sets, the average specificity is 13.2% (standard deviation, 14.2%), which indicates that the 50% specificity found by our gene list is significantly higher (z value, 2.59; two-tailed P = .0096) than that of the random data sets. Finally, because ER status has also been associated with tumor relapse to bone,1 we analyzed the performance of ER status to predict tumor relapse to bone in our test set. Of the 27 ER-positive tumors, 20 relapsed to the bone; of eight ER-negative tumors, three relapsed to the bone (74% sensitivity and 63% specificity). Thus, ER status performs less well compared with the bone signature to predict a tumor relapse to bone.
Annotation of Biologic Processes and Pathways to Bone Metastasis
We next studied the 142 genes from the combined list that were annotated in Gene Ontology and determined whether certain Gene Ontology descriptions were over-represented in the merged SAM/Kang list compared with all genes present on the U133a chip.13,15 Over-represented annotations point to biological processes, which potentially are linked to the site of relapse. For example, the description "extracellular" was linked to 21 of the 142 (14.8%) genes from the bone marker list, whereas 1,350 of 16,367 genes (8.2%) of the U133a chip were annotated to this description. This means "extracellular" is 1.8 times over-represented (P = .004;  2 distribution) in the bone relapse list. Other examples are "cell adhesion" (17 genes; P = .0005) and "cell organization and biogenesis" (22 genes; P = 1.4 x 10–5), which were found to be 2.2 and 2.4 times over-represented, respectively. In addition, "immune response" was significant (P = 6.9 x 10–5), but in contrast to the above-mentioned descriptions, the genes linked to "immune response" originated predominantly from the Kang list.
The biology of organ-specific metastasis is poorly understood. In this study we have searched for genes expressed in the primary tumors that preferentially relapse to bone. The SAM analysis yielded 69 differentially expressed genes between tumors relapsing to bone versus tumors relapsing elsewhere in the body. The trefoil protein encoding genes TFF1 and TFF3 were the highest ranking genes in the SAM analysis and validation of TFF1 by quantitative RT-PCR in a completely independent cohort confirmed that TFF1 expression was positively correlated with a tumor relapse to bone, suggesting the importance of these proteins in this process. TFF3 is found overexpressed in some primary and metastatic prostate cancers,16 whereas recent publications indicate that TFF1 induces cellular invasion of kidney and colon cancer cells.17,18 Thus, it is possible that in breast cancer, TFF1, using a similar mechanism, contributes selectively to invasion of bone. The fact that TFF1 may contribute to tumor relapse to bone is underscored by its abundant presence in breast cancer micrometastases.19 However, because the TFF1 protein expression is usually lost during progression of some GI tract adenocarcinomas,20 it is intriguing to ascertain if TFF3 is of importance to consolidate breast cancer metastasis. Other data suggest that TFF peptides can dimerize with cysteine-rich molecules,20 and it is interesting in this regard that we identified cysteine-rich intestinal protein 1 (CRIP1) as one of the top five genes in our SAM analysis, which thus might be an interacting partner for TFF1 required for preferential relapse to bone. The expression of these two genes is highly correlated in the expression array samples (n = 286; Pearson r = 0.59; P < 10–10), suggesting common regulation or biology. As hypothesized in the seed and soil theory of Paget,21 the molecular interaction of circulating cancer cells with the microenvironment of the bone matrix plays an important role in bone metastasis and depends on cell-to-cell interactions involving cell adhesion proteins between marrow stromal cells and cancer cells, and other types of communication between bone and cancer cells.22 In line with this view, we indeed found through our Gene Ontology analysis an over-representation of genes linked to "cell adhesion." Of the 17 genes linked to this description, six genes were found to be more highly expressed in bone-metastasizing cells/tumors: MCAM, PTK7, CTGF (Kang list), RND1, TSPAN1, and ANXA9 (SAM). These genes may thus participate in adhesion processes required for the colonization of bone. In addition, we identified through the KEGG pathway analysis that the FGFR-MAPK signaling pathway is over-represented in tumors spreading to bone. To have enough power in this analysis, we included the bone metastasis gene list of Kang et al6 derived from an in vitro cell model preferentially relapsing to bone in nude mice. With the exception of the BENE gene, these two lists of significant genes had no overlap. This may be due to the obvious differences in samples used (ie, tumors including stromal components from a mixed ER population versus an ER-negative cell line without stroma). However, it is reassuring that the clinical and in vitro studies point to a common pathway. The exact effect of the FGF signaling pathway with respect to bone metastasis remains to be elucidated. However, it is intriguing that the growth factors FGF-1 and FGF-2 are stored in mineralized bone matrices,23 which is reminiscent of the chemoattraction displayed by some organs (including bone) that stimulates circulating cancer cells to colonize these organs. Heparanase, recently linked to bone metastasis of myeloma cells,24 may participate in this process, given that it can degrade and thereby release heparan sulfate–immobilized FGF. It is obvious that additional research using specific inhibitors or RNA interference–mediated knock-down experiments are needed to substantiate the role of the identified pathway on the ability to form bone metastasis in breast cancer. The PAM analysis identified a 31-gene profile that was able to identify all samples with tumor relapse to bone in the testing set with a 50% specificity. The bone profile performed better than randomly chosen gene sets, but we are prudent in presenting our PAM list as a diagnostic profile for therapy decisions; we realize that a much larger confirmatory study is essential to validate its performance. If the bone profile is sufficiently validated, the use of a high-risk profile (eg, our 76-gene profile5 to identify patients at high risk for a distant recurrence), in combination with the bone profile would identify patients eligible for therapy with bisphosphonates. At present, the bone profile leads to overtreatment of 50% of the patients who will not experience a relapse to bone. Although it is advantageous to avoid overtreatment (ie, to increase the profile’s specificity), the undesired overtreatment of patients is probably outweighed by the fact that bisphosphonate therapy is well tolerated.25 Kang et al6 also analyzed their bone profile in 25 primary breast tumors.7 Using hierarchical clustering, two groups could be distinguished. One of the groups had 10 of 15 samples with a tumor relapse to bone (67% sensitivity), whereas the other group showed eight of 10 samples from areas with nonbone relapse (80% specificity). An analogous study describing a lung metastasis profile for breast cancer showed promising results in clinical samples.26 Future validation studies are needed to confirm these important findings. We recently published a 76-gene profile5 that can identify breast cancer patients at high risk for a distant metastasis within 5 years. We were not able to correlate this high-risk profile to the site of relapse, nor did we find an association between the time until distant metastasis and site of relapse in our extensive patient database. Translating the high-risk profile and the bone profile to the general characteristics of a tumor, we conclude that aggressiveness of a tumor (ability to metastasize) is distinct from the tumor’s ability to home and proliferate in an organ-specific manner. In this respect, our study performed on human tumor specimens is completely in line with observations on human cell line subclones xenographed in animal models.6,7 We envision the use of the 76-gene profile5 to identify patients at high risk for a recurrence, whereas the bone profile can further subgroup the patients who will develop a tumor relapse to bone. This subgroup is clinically relevant because these patients may benefit from additional therapy with bisphosphonates.8 In conclusion, we have identified a set of genes including the trefoil protein encoding genes TFF1 and TFF3, a biological property (cell adhesion), and a signaling pathway (FGFR-MAPK) over-represented in primary breast cancers that preferentially relapse to the bone. We hypothesize that the indicated biologic features have an important role in bone metastasis, and establishment of the modus operandi may greatly benefit the large group of breast cancer patients with a relapse to the bone.
The gene list of interest as well as the complete Affymetrix U133 gene list was linked to the Gene Ontology (GO) data using the Unigene cluster number as a linking pin. See Smid and Dorssers (Smid M, Dorssers LC: GO-Mapper: Functional analysis of gene expression data using the expression level as a score to evaluate GO terms. Bioinformatics 20:2618-2625, 2004) for details about the linking procedure. The Perl scripts described therein were modified so that an output file is generated that contains the number and identity of genes linked to a specific GO description. By merging the data from both the gene list of interest and complete array list, the relative frequencies can be calculated. For example, the description "extracellular" was linked to 21 of the 142 (14.8%) genes from the bone marker list, whereas 1,350 of 16,367 genes (8.2%) of the complete U133A chip were annotated to this description. This means "extracellular" is 1.8 times over-represented.
For the statistical evaluation we made use of the
All descriptions listed in the table below comply with the minimum parameters regarding expected frequencies when using the  2 distribution. Over-represented descriptions were selected when P < .01 (Table A2).
Figure A1 (parts A and B) gives the GO hierarchy of GO descriptions associated with bone relapse (descriptions that were not significant are grayed out; Fig A1).
We chose the low-level descriptions as the most informative (ie, least generic) annotation. Thus, we considered "immune response," "cell adhesion," "cell organization and biogenesis," and "extracellular" as significantly over-represented descriptions.
The genes that belong to the descriptions mentioned in the main text are listed in Tables A3
Although all authors completed the disclosure declaration, the following authors or their immediate family members indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. For a detailed description of the disclosure categories, or for more information about ASCO’s conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.
Dollar Amount Codes (A) < $10,000 (B) $10,000-99,999 (C)
Supported in part by the Netherlands Genomics Initiative/Netherlands Organisation for Scientific Research. This organization had no role in study design, the collection or analysis of data, writing the article, or in decisions relating to publication. Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
1. James JJ, Evans AJ, Pinder SE, et al: Bone metastases from breast carcinoma: Histopathological-radiological correlations and prognostic features. Br J Cancer 89:660-665, 2003[CrossRef][Medline] 2. Coleman RE: Skeletal complications of malignancy. Cancer 80:1588-1594, 1997[CrossRef][Medline] 3. Koenders PG, Beex LV, Langens R, et al: Steroid hormone receptor activity of primary human breast cancer and pattern of first metastasis: The Breast Cancer Study Group. Breast Cancer Res Treat 18:27-32, 1991[CrossRef][Medline] 4. Roodman GD: Mechanisms of bone metastasis. N Engl J Med 350:1655-1664, 2004 5. Wang Y, Klijn JG, Zhang Y, et al: Gene-expression profiles to predict distant metastasis of lymph-node-negative primary breast cancer. Lancet 365:671-679, 2005[Medline] 6. Kang Y, Siegel PM, Shu W, et al: A multigenic program mediating breast cancer metastasis to bone. Cancer Cell 3:537-549, 2003[CrossRef][Medline] 7. Minn AJ, Kang Y, Serganova I, et al: Distinct organ-specific metastatic potential of individual breast cancer cells and primary tumors. J Clin Invest 115:44-55, 2005[CrossRef][Medline] 8. Lipton A: Management of bone metastases in breast cancer. Curr Treat Options Oncol 6:161-171, 2005[Medline] 9. Tusher VG, Tibshirani R, Chu G: Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A 98:5116-5121, 2001 10. Tibshirani R, Hastie T, Narasimhan B, et al: Diagnosis of multiple cancer types by shrunken centroids of gene expression. Proc Natl Acad Sci U S A 99:6567-6572, 2002 11. Martens JW, Nimmrich I, Koenig T, et al: Association of DNA methylation of phosphoserine aminotransferase with response to endocrine therapy in patients with recurrent breast cancer. Cancer Res 65:4101-4117, 2005 12. Foekens JA, van Putten WL, Portengen H, et al: Prognostic value of PS2 and cathepsin D in 710 human primary breast tumors: Multivariate analysis. J Clin Oncol 11:899-908, 1993 13. Ashburner M, Ball CA, Blake JA, et al: Gene ontology: Tool for the unification of biology—The Gene Ontology Consortium. Nat Genet 25:25-29, 2000[CrossRef][Medline] 14. Kanehisa M, Goto S: KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic Acids Res 28:27-30, 2000 15. Smid M, Dorssers LC: GO-Mapper: Functional analysis of gene expression data using the expression level as a score to evaluate Gene Ontology terms. Bioinformatics 20:2618-2625, 2004 16. Garraway IP, Seligson D, Said J, et al: Trefoil factor 3 is overexpressed in human prostate cancer. Prostate 61:209-214, 2004[CrossRef][Medline] 17. Rodrigues S, Nguyen QD, Faivre S, et al: Activation of cellular invasion by trefoil peptides and src is mediated by cyclooxygenase- and thromboxane A2 receptor-dependent signaling pathways. FASEB J 15:1517-1528, 2001 18. Rodrigues S, Attoub S, Nguyen QD, et al: Selective abrogation of the proinvasive activity of the trefoil peptides pS2 and spasmolytic polypeptide by disruption of the EGF receptor signaling pathways in kidney and colonic cancer cells. Oncogene 22:4488-4497, 2003[CrossRef][Medline] 19. Mikhitarian K, Gillanders WE, Almeida JS, et al: An innovative microarray strategy identities informative molecular markers for the detection of micrometastatic breast cancer. Clin Cancer Res 11:3697-3704, 2005 20. Emami S, Rodrigues S, Rodrigue CM, et al: Trefoil factor family (TFF) peptides and cancer progression. Peptides 25:885-898, 2004[CrossRef][Medline] 21. Paget S: The distribution of secondary growths in cancer of the breast. 1889. Cancer Metastasis Rev 8:98-101, 1989[Medline] 22. Yoneda T, Hiraga T: Crosstalk between cancer cells and bone microenvironment in bone metastasis. Biochem Biophys Res Commun 328:679-687, 2005[CrossRef][Medline] 23. Hauschka PV, Mavrakos AE, Iafrati MD, et al: Growth factors in bone matrix: Isolation of multiple types by affinity chromatography on heparin-Sepharose. J Biol Chem 261:12665-12674, 1986 24. Yang Y, Macleod V, Bendre M, et al: Heparanase promotes the spontaneous metastasis of myeloma cells to bone. Blood 105:1303-1309, 2005 25. Li EC, Davis LE: Zoledronic acid: A new parenteral bisphosphonate. Clin Ther 25:2669-2708, 2003[CrossRef][Medline] 26. Minn AJ, Gupta GP, Siegel PM, et al: Genes that mediate breast cancer metastasis to lung. Nature 436:518-524, 2005[CrossRef][Medline] Submitted August 18, 2005; accepted December 28, 2005.
Related Editorials
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2006 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
$100,000 (N/R) Not Required
