CLINICAL RESEARCH: HEART FAILURE
Identification of a Common Gene Expression Signature in Dilated Cardiomyopathy Across Independent Microarray Studies
Andreas S. Barth, MD*,*,
Ruprecht Kuner, PhD ,
Andreas Buness, MSc,
Markus Ruschhaupt, MSc ,,
Sylvia Merk, DVM,||,
Ludwig Zwermann, MD*,
Stefan Kääb, MD*,
Eckart Kreuzer, MD ,
Gerhard Steinbeck, MD*,
Ulrich Mansmann, PhD,
Annemarie Poustka, PhD,
Michael Nabauer, MD* and
Holger Sültmann, PhD
* Department of Medicine I, University Hospital Grosshadern, Munich, Germany
Department of Cardiac Surgery, University Hospital Grosshadern, Munich, Germany
Department of Medical Informatics, Biometrics and Epidemiology, Ludwig-Maximilians-University, Munich, Germany
Division of Molecular Genome Analysis, German Cancer Research Center, Heidelberg, Germany
|| Department of Medical Informatics and Biomathematics, University of Münster, Münster, Germany
Manuscript received December 18, 2005;
revised manuscript received April 14, 2006,
accepted June 6, 2006.
* Reprint requests and correspondence: Dr. Andreas S. Barth, Department of Medicine I, University Hospital Grosshadern, Munich, Marchioninistrasse 15, 81377 Munich, Germany (Email: andreas.barth{at}med.uni-muenchen.de).
 |
Abstract
|
|---|
OBJECTIVES: This study was designed to identify a common gene expression signature in dilated cardiomyopathy (DCM) across different microarray studies.
BACKGROUND: Dilated cardiomyopathy is a common cause of heart failure in Western countries. Although gene expression arrays have emerged as a powerful tool for delineating complex disease patterns, differences in platform technology, tissue heterogeneity, and small sample sizes obscure the underlying pathophysiologic events and hamper a comprehensive interpretation of different microarray studies in heart failure.
METHODS: We accounted for tissue heterogeneity and technical aspects by performing 2 genome-wide expression studies based on cDNA and short-oligonucleotide microarray platforms which comprised independent septal and left ventricular tissue samples from nonfailing (NF) (n = 20) and DCM (n = 20) hearts.
RESULTS: Concordant results emerged for major gene ontology classes between cDNA and oligonucleotide microarrays. Notably, immune response processes displayed the most pronounced down-regulation on both microarray types, linking this functional gene class to the pathogenesis of end-stage DCM. Furthermore, a robust set of 27 genes was identified that classified DCM and NF samples with >90% accuracy in a total of 108 myocardial samples from our cDNA and oligonucleotide microarray studies as well as 2 publicly available datasets.
CONCLUSIONS: For the first time, independent microarray datasets pointed to significant involvement of immune response processes in end-stage DCM. Moreover, based on 4 independent microarray datasets, we present a robust gene expression signature of DCM, encouraging future prospective studies for the implementation of disease biomarkers in the management of patients with heart failure.
|
Abbreviations and Acronyms
| | DCM = dilated cardiomyopathy | | GO = Gene Ontology | | ICM = ischemic cardiomyopathy | | NF = nonfailing |
|
Dilated cardiomyopathy (DCM) is characterized by dilatation and impaired contraction of 1 or both ventricles in the absence of significant coronary artery disease. The incidence of DCM has been estimated to be 5 to 8 cases per 100,000 individuals, with a prevalence of 36 per 100,000 (1). Thus, DCM is a leading cause of heart failure and cardiac transplantation in Western countries (2). The high morbidity and mortality associated with DCM underscore the need for a better understanding of the underlying molecular events leading to heart failure in DCM. So far, the natriuretic peptides are the best characterized markers for diagnosing and managing heart failure. However, there is significant heterogeneity in expression levels of these molecular disease markers that is not explained by left ventricular function alone (3–5). Transcriptional signature analysis is a powerful technique capable of identifying new molecular targets which could ultimately become important for diagnosis and therapy of heart failure. Yet, differences in platform technologies, experimental design, and the biological heterogeneity associated with the use of human tissue samples form obstacles to the successful comparison and integration of results obtained by different microarray studies in heart failure. In addition, substantial regional variation in gene expression exists in mammalian myocardium including atrium, ventricle and septum or left and right side of the heart (6–9). These differences make it difficult to determine which transcripts are related to DCM as opposed to other sources of biological variability.
The aim of the present study was 2-fold: First, we wanted to identify common genes and shared biological processes based on Gene Ontology (GO) (10) across 2 independent microarray studies in human DCM to minimize over-interpretation of either dataset. Biological variability was accounted for by examining 20 DCM and 20 nonfailing (NF) myocardial samples from different regions of human myocardium (septum and left ventricle of different transmural origin). Our second goal aimed at establishing a robust set of genes of DCM and NF hearts for the 2 present and 2 additional, publicly available, microarray datasets. As a result, we present a set of 27 genes which classified 108 NF and DCM samples in 4 independent microarray datasets with >90% accuracy.
|
Methods
|
|---|
Study design.
We performed 2 microarray studies with a total of 40 patient samples. The cDNA microarray study (dataset A) was based on 28 septal myocardial samples obtained from 13 DCM hearts at the time of transplantation and 15 NF donor hearts that were not transplanted because of palpable coronary calcifications. The latter patient group was not known to have any history of overt cardiovascular disease. For our oligonucleotide microarray study (dataset B), 12 independent subendocardial left ventricular samples were collected from 7 DCM patients and 5 NF donors. Detailed patient characteristics are listed in (see ). All transplanted patients gave written informed consent. The investigation was approved by the Institutional Review Board.
After excision, all tissue specimens were frozen in liquid nitrogen and stored at –80°C. RNA isolation, sample preparation, labeling, and hybridization to RZPD Unigene 3.1 cDNA (37.5 K) and to Affymetrix U133A (22.2 K) arrays were carried out as described previously (6,11,12). The original data files for datasets A and B were deposited in the Gene Expression Omnibus database (GEO) (13) and are accessible through GEO series accession numbers GSE3585 and GSE3586.
For the classification and the verification of the classifier gene set, we included 2 additional studies, for which data were publicly available. Dataset C consists of 6 NF, 21 DCM, and 10 ischemic cardiomyopathy (ICM) samples hybridized to Affymetrix HG-U133A arrays (14). Normalized gene expression data were downloaded from GEO (GSE1869
[NCBI GEO]
). Dataset D is available online through a program for genomic application funded by the National Heart, Lung, and Blood Institute and consists of 14 NF, 27 DCM, and 32 ICM samples hybridized to Affymetrix HG-U133 2.0-plus arrays (15). In summary, we used 68 DCM and 40 NF samples from 4 independent studies for classification.
Data extraction and statistical analysis.
Preprocessing and most of the statistical analysis were done using R (16) and Bioconductor (17). After quality control, all cDNA microarray data were normalized using arrayMagic (18) and "VSN" (19). Normalized data were filtered with respect to signal intensity. Quality of the HG-U133A arrays was assured by controlling for dynamic range, perfect match saturation, pixel noise, grid misalignment, and signal-to-noise ratio. Microarray data of all samples in microarray study B were normalized in common using robust multi-array average (RMA) (20) implemented in Bioconductor’s "affy" package. Probe sets with "absent" calls in more than 50% of tissue samples in either group (NF and DCM) were excluded.
To determine differentially expressed genes, 2-class unpaired significance analysis of microarrays (SAM) (21) was applied in both studies. Differences in gene expression were regarded as statistically relevant if a false discovery rate (FDR) of q < 0.05 and a fold-change of  1.2 were achieved.
Mapping of transcripts between cDNA clones and Affymetrix probe sets was achieved by means of the MatchMiner software tool (22). Functional annotation of differentially expressed genes was based on hierarchical system of GO domains "cellular component," "biological process," and "molecular function." Over-representation of specific GO classes in a gene set was statistically analyzed by "FatiGO" (23).
Expression values of selected transcripts were validated by quantitative real-time polymerase chain reaction (PCR) (ABI Prism 7900HT Sequence Detection System; Applied Biosystems, Weiterstadt, Germany). A detailed list of genes examined by RT-PCR including protocols used for RT-PCR is given in the supplemental Materials and Methods section (see ).
Identification of a gene expression signature for DCM.
Prediction analysis for microarrays (PAM) (24) was used for classification. The ability to correctly classify the status of DCM and NF samples was assessed by complete cross-validation implemented in the Bioconductor package "MCRestimate" (25). First, the samples in every study were randomly divided into equally sized subsets. In each following step, 1 subset was left aside and the classifier (filtering and PAM) was built on the remaining samples (training set). The status (NF vs. DCM) of the left-out samples was predicted and compared with the clinically diagnosed status. Optimization of the PAM parameter and of the number of genes remaining after variance filtering was achieved through a second cross-validation within each training set. To estimate the variability of the cross-validation result based on different sample compositions of the training set, the procedure was repeated 50 times. A sample was called "misclassified" if it was incorrectly classified in more than half of all cross-validations.
|
Results
|
|---|
Common changes of biological processes in DCM.
To identify differentially expressed genes, we analyzed our 2 experimental studies, which included septal myocardial tissue samples from 13 DCM and 15 NF hearts (dataset A) and left ventricular samples from 7 DCM and 5 NF hearts (dataset B) by SAM analysis. In dataset A, 1,353 transcripts were up-regulated and 384 were down-regulated in DCM. In dataset B, 399 transcripts were up-regulated and 75 transcripts were down-regulated in DCM (, see ). In both studies, up-regulation was about 4 to 5 times more common than down-regulation, indicating a net transcriptional activation in heart failure. Overall, 76 transcripts were found to be consistently deregulated in both studies, representing  16% overlap at the single gene level between both microarray studies (). These transcripts included known marker genes of heart failure, such as pro-brain natriuretic peptide (pro-BNP) (26), and chemokine (C-C motif) ligand 2 (CCL2) (27), but also many genes previously not associated with cardiomyopathies. Validation of microarray expression values was done by using quantitative real-time PCR. As a result, we found a strong correlation for the expression ratios between arrays and quantitative PCR for 11 of 12 differentially expressed genes analyzed in dataset A (, see ).
To gain a comprehensive insight into the biological processes associated with DCM, we related differentially expressed genes to their respective GO classes. Thereby, we were able to identify specific biological processes which were consistently enriched in up- or down-regulated transcripts of both studies (, see ). For example, both studies showed a marked up-regulation of transcripts involved in protein biosynthesis in DCM. However, it is interesting to note that this functional GO class comprises qualitatively different genes in the 2 studies. Whereas the cDNA microarray study (dataset A) detected many elongation factors, ribosomal transcripts were more frequently recognized in the Affymetrix study (dataset B) (, see ).
The biological processes "immune" and "inflammatory response" displayed the most significant changes in the group of down-regulated genes in DCM. These functional gene classes included components of the complement system (C1QB, C1QR1, C1R, C3), chemokines (CCL2, CCL11, CCL18), interferon-induced genes (IFI27, IFI30, IFITM1, IFITM3, STAT3), calgranulins (S100A8, S100A9), and leukocyte antigens (CD14, CD53, CD163), suggesting a profound deregulation of the immune system in DCM (Fig. 1). In accordance with down-regulation of immune response genes in DCM, the functional class of "chemokine activity" displayed the most prominent down-regulation specified by level 6 of GO category "molecular function."
View larger version (31K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1 Functional analysis based on Gene Ontology for selected gene classes comparing up-regulated genes in DCM (yellow bars) and down-regulated genes (green bars) in dilated cardiomyopathy (DCM) of datasets A (open bars) and B (striped bars). Differences in gene classes marked by an asterisk were statistically significant (Fisher exact test; p < 0.05) according to "FatiGO" (23). Results for cellular component, biological process, and molecular function are based on (see ).
|
|
Consistent with prominent structural remodeling in end-stage DCM, many deregulated transcripts were related to extracellular matrix composition and turnover. Notably, up-regulation of collagen transcripts (COL5A1, COL8A1) and the procollagen C-endopeptidase enhancer 2 (PCOLCE2) which binds to type I procollagen and potentiates its cleavage by procollagen C-proteinases, was observed. In addition, extracellular matrix protein 2 (a member of the small leucine rich proteoglycans (SLRP), important for collagen fibrillogenesis), asporin, and most other members of the SLRP family were found to be up-regulated as well (decorin, lumican, biglycan, fibromodulin, osteoglycin, and osteomodulin), highlighting their importance in extracellular remodeling. Furthermore, we noted prominent up-regulation of genes coding for Z-disc components, including caldesmon 1, sarcospan, sarcoglycan epsilon, utrophin, spectrin, titin, vinculin, sarcoglycan D and G, alpha-actinin, LIM-domain binding 3, and alpha-2-capping protein. The Z-disc is thought to act as a sensor, linking biomechanical forces to the activation of stress pathways (28). In this sense, we were able to corroborate the desensitization of beta-adrenergic signaling by gene expression analysis with down-regulation of adrenergic, beta-1-receptor in end-stage DCM (, see ). Profound changes in signal transduction were also reflected in the down-regulation of the GO class "integral to plasma membrane" specified by level 6 of GO category "cellular component." With regard to the deregulation of important signaling pathways, we found transcriptional repression of the oncostatin M receptor (OSMR), anti-apoptotic gene BCL2L1, and signal transducer STAT3, which are involved in the protection of the myocardium from heart failure via the JAK-STAT pathway. Ultimately, the balance between pro- and anti-apoptotic programs may determine if relevant loss of myocytes occurs. In line with this notion, we noted up-regulation of anti- (FGF1, DSIPI) and pro-apoptotic transcripts (CCL2, BCLAF1, FOXO3A).
Identification of a gene expression signature for DCM.
The second goal of our analysis was to identify a specific set of transcripts that could reliably classify DCM and NF samples. We approached this issue by performing the PAM classification method in 4 independent microarray studies. Very low misclassification rates were found in our 2 studies (datasets A and B) and in dataset D for the classification of NF vs. DCM samples (Fig. 2). One out of 12 samples was misclassified in dataset B. Likewise, datasets A and D showed similar results, with 1 out of 28 and 3 out of 41 misclassified samples, respectively. In contrast, the classification algorithm did not show any predictive power in dataset C. This was unexpected, because the expression levels of established molecular cardiomyopathy markers, including pro-BNA or pro-ANP, suggested a clear separation into NF and failing ventricular samples (Fig. 3).
View larger version (39K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2 Prediction analysis for microarrays (PAM) classification. In the first step, PAM classification was applied to all 4 datasets separately. Very low misclassification rates were found in datasets A, B, and D for the classification of nonfailing (NF) versus dilated cardiomyopathy (DCM) samples, whereas the classification algorithm did not show any power in dataset C. In the second step, we repeated the procedure with the smallest gene signature obtained from dataset B (31 probe sets which correspond to 27 transcripts), now achieving more than 90% accuracy for classifying DCM and NF samples across all studies, including dataset C.
|
|
View larger version (16K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3 Mean expression ± SEM of pro-brain natriuretic peptide (NPPB) in nonfailing (NF) (black bars) and dilated cardiomyopathy (DCM) samples (red bars) in datasets A to D. Statistical comparison was carried out by Student t test.
|
|
The smallest number of probe sets used for classification was found in dataset B, with a median of 5 probe sets comprising 31 different probe sets and a median absolute deviation of 2.9. Therefore, we evaluated the ability of this set of 31 probe sets coding for 27 genes (Table 1) to correctly classify DCM and NF samples of the remaining 3 studies. First, we reduced datasets C and D to the 31 probe sets that were used for classification in dataset B. Next, we used PAM without filtering on the reduced set for classification and estimated the prediction power by performing a complete classification procedure. For dataset A, we first determined the presence of 17 out of 27 genes on the cDNA microarray and proceeded with the classification as previously described. In summary, we found that a small set of 27 genes was sufficient to classify DCM and NF hearts across all 4 independent studies with more than 90% accuracy (Fig. 2). Remarkably, this comprehensive gene set was able to classify DCM and NF samples in dataset C, for which the PAM method had initially failed.
The 27-gene signature included known marker genes of heart failure like pro-BNP, pro-atrial natriuretic peptide (pro-ANP), corin (which converts pro-ANP to biologically active ANP), transcripts encoding for sarcomer structure proteins (MYH6, MYH10), apoptotic processes (CCL2, PHLDA1, SNCA), cell growth (FRZB, SFRP4, SPOCK, CTGF), and cell cycle control (G0S2, ETV5, RARRES1). Notably, the selection of individual known marker genes such as pro-BNP alone was not sufficient to classify the DCM cases because of the heterogeneous gene expression across all 108 myocardial samples (Fig. 3).
Because several previously described genes for heart failure were part of the classifier gene set, its validity might well hold for heart failure in general, irrespective of etiology. We tested this hypothesis by PAM analysis for ICM and NF samples included in datasets C (ICM: n = 10; NF: n = 6) and D (ICM: n = 32; NF: n = 14) and found that this gene set classified more than 90% of these samples correctly as well (data not shown).
|
Discussion
|
|---|
Different etiologies and duration of dilated cardiomyopathy, differences in age, gender, and medications, as well as individual course of the disease contribute to the variability of gene expression data. In addition, it is very difficult to obtain true "non-failing" human ventricular tissue, because donor hearts may have been exposed to varying degrees of hypoxia or hemodynamic stress which are known to be potent inducers of chemokine and BNP gene expression (29). In the present work, we considered these variables by integrating independent microarray studies from a large number of failing and non-failing hearts to identify common transcripts and biological processes involved in the pathogenesis of DCM and to define a robust, common denominator for DCM and NF hearts. Based on the smallest gene set for classification from dataset B, the PAM method classified all 4 human heart failure datasets with low misclassification rates. Notably, despite the large variation of gene expression values for single genes, the classifier as a whole is highly valuable to distinguish DCM and NF samples. This supports the usefulness of this molecular approach for future diagnostic applications. In addition, the classificator gene set based on DCM and NF hearts also achieved a similarly high accuracy of classification in ICM as in DCM samples, suggesting that this gene set could be representative of molecular changes of heart failure in general.
Of note, the classificator based on dataset B performed as well in datasets A and D as if one used classificators generated from these 2 datasets alone. What is more, the gene signature from dataset B was also able to accurately discriminate NF and DCM samples in dataset C. As noted by the authors of dataset C, differences in gene expression were greater between left ventricular assist-device (LVAD) and non-LVAD hearts in the DCM group than between DCM and NF samples (14). This peculiarity might impede the PAM approach for identifying a useful classifier between DCM and NF within this dataset itself.
The classifier gene signature can be grouped into different functional sets with respect to the pathogenesis of DCM. Up-regulation of the cardiomyopathy markers pro-ANP and pro-BNP is well established in heart failure (26) and mediated by neurohormonal dysregulation (30). Activation of pro-fibrotic stress hormone pathways leads to prominent structural remodeling in DCM, exemplified by deregulation of genes coding for sarcomer structure and extracellular matrix proteins such as myosin 6 and 10, asporin, PCOLCE2, kelch-like 3 (KLHL3), and adipocyte enhancer binding protein 1 (AEBP1). In addition to pro-BNP, further transcripts of this set of classifier genes, including the transcription factor ZBTB16 (31), the connective tissue growth factor CTGF (32), and the chemokine CCL2 (33), characterize important targets of the renin-angiotensin system in failing myocardium, because they all have been shown to be regulated by angiotensin-II.
Other transcripts of this classifier gene set belong to apoptotic (PHLDA1, SNCA, CCL2) and cell growth (FRZB, SFRP4, SPOCK, CTGF) processes. Of note, the 2 genes coding for frizzled-related protein (FRZB) and secreted frizzled-related protein 4 (SFRP4) are members of the Wnt signaling pathway, implicated in wound healing and regeneration of heart failure (34,35). Especially, the expression of the Wnt antagonist SFRP4 is associated with myocyte apoptosis in overload-induced heart failure (34).
Furthermore, the genes coding for the complement factor H-related 3 (CFHL3), ficolin 3 (FCN3), chemokine ligand 2 (CCL2), and calgranulin A (S100A8) are related to stress and immune response. Remarkably, the GO classes "immune response" and "inflammatory response" showed the most significant changes of all biological processes in our 2 datasets, A and B. Given that DCM has a very heterogeneous etiology, it is plausible that a subgroup of DCM represents post-infectious autoimmune disease, especially in individuals with genetic susceptibility. However, independent experimental models of cardiomyopathy suggest that cardiac remodeling itself is able to trigger immune response (36,37). Chemokine ligand 2 is a prominent member of the broader functional group of immune and inflammatory processes and was found to be down-regulated in both datasets A and B. This chemokine, capable of interacting with tumor necrosis factor alpha and interleukin-6–related pathways, has been localized to the cardiomyocyte compartment by immunohistochemistry (27). It promotes attraction and invasion of activated leukocytes into the failing myocardium but is also involved in shaping the extracellular matrix by modulating the activity of matrix metalloproteinases and collagen turnover (38) as well as cell proliferation and induction of apoptosis (27). Down-regulation of CCL2 transcripts in end-stage heart failure may therefore represent an adaptive mechanism to promote cell survival. Of note, additional chemokines like CCL11 and CCL18 also were found to be down-regulated in Affymetrix and Unigene arrays, respectively.
Given the limited availability of human myocardial samples, future studies will need to correlate gene expression patterns in myocardium to tissue specimens more readily available from individual patients and suitable as myocardial surrogates (possibly peripheral blood leukocytes). In addition to the known cardiomyopathy markers pro-ANP and pro-BNP, we identified further genes involved in the natriuretic system and immune response processes which may be promising candidates for disease biomarkers. In this respect, the identification of a robust molecular signature of DCM across different microarray platforms and independent studies enables careful "hypothesis driven" validation studies of biomarkers of heart failure and testing their clinical utility.
|
Appendix
|
|---|
For an extended version of the Materials and Methods and supplemental tables, please see the online version of this article.
|
Footnotes
|
|---|
Supported by a grant from the German Federal Ministry for Education and Research supporting the German National Genome Research Network (BMBF-grant 01GS0109, NGFN grant 01GR0101, and NGFN grant 01GR0459) and a grant from the Deutsche Forschungsgemeinschaft (DFG grant BA 3341/1-1). Affymetrix data were generated and analyzed in the framework of a research collaboration with Aventis Pharma Deutschland. Drs. Barth, Kuner, Nabauer, and Sültmann contributed equally to this work.
|
References
|
|---|
1. Dec GW, Fuster V. Idiopathic dilated cardiomyopathy N Engl J Med 1994;331:1564-1575.[Free Full Text]2. Roger VL, Weston SA, Redfield MM, et al. Trends in heart failure incidence and survival in a community-based population JAMA 2004;292:344-350.[Abstract/Free Full Text] 3. Hwang JJ, Allen PD, Tseng GC, et al. Microarray gene expression profiles in dilated and hypertrophic cardiomyopathic end-stage heart failure Physiol Genomics 2002;10:31-44.[Abstract/Free Full Text] 4. Tan FL, Moravec CS, Li J, et al. The gene expression fingerprint of human heart failure Proc Natl Acad Sci U S A 2002;99:11387-11392.[Abstract/Free Full Text] 5. Yung CK, Halperin VL, Tomaselli GF, et al. Gene expression profiles in end-stage human idiopathic dilated cardiomyopathy: altered expression of apoptotic and cytoskeletal genes Genomics 2004;83:281-297.[CrossRef][Web of Science][Medline] 6. Barth AS, Merk S, Arnoldi E, et al. Functional profiling of human atrial and ventricular gene expression Pflügers Archiv 2005;450:201-208.[CrossRef][Web of Science][Medline] 7. Nabauer M, Beuckelmann DJ, Uberfuhr P, et al. Regional differences in current density and rate-dependent properties of the transient outward current in subepicardial and subendocardial myocytes of human left ventricle Circulation 1996;93:168-177.[Abstract/Free Full Text] 8. Ramakers C, Stengl M, Spatjens RLHM, et al. Molecular and electrical characterization of the canine cardiac ventricular septum J Mol Cell Cardiol 2005;38:153-161.[CrossRef][Web of Science][Medline] 9. Tabibiazar R, Wagner RA, Liao A, Quertermous T. Transcriptional profiling of the heart reveals chamber-specific gene expression patterns Circ Res 2003;93:1193-1201.[Abstract/Free Full Text] 10. Ashburner M, Ball CA, Blake JA, et al. Gene Ontology Consortium. Gene ontology: tool for the unification of biology Nat Genet 2000;25:25-29.[CrossRef][Web of Science][Medline] 11. Sultmann H, von Heydebreck A, Huber W, et al. Gene expression in kidney cancer is associated with cytogenetic abnormalities, metastasis formation, and patient survival Clin Cancer Res 2005;11:646-655.[Abstract/Free Full Text] 12. Barth AS, Merk S, Arnoldi E, et al. Reprogramming of the human atrial transcriptome in permanent atrial fibrillation—expression of a ventricular-like genomic signature Circ Res 2005;96:1022-1029.[Abstract/Free Full Text] 13. Gene Expression Omnibus. Available at: http://www.ncbi.nlm.nih.gov/geo. Accessed November 9, 2005.. 14. Kittleson MM, Minhas KM, Irizarry RA, et al. Gene expression analysis of ischemic and nonischemic cardiomyopathy: shared and distinct genes in the development of heart failure Physiol Genomics 2005;21:299-307.[Abstract/Free Full Text] 15. NHLBI Program for Genomic Applications, Harvard Medical School. Genomics of cardiovascular development, adaptation, and remodeling. Available at: http://www.cardiogenomics.org. Accessed October 16, 2004.. 16. The R Project for Statistical Computing. Available at: http://www.r-project.org. Accessed January 1, 2005.. 17. Bioconductor. Available at: http://www.bioconductor.org. Accessed January 1, 2005.. 18. Buness A, Huber W, Steiner K, et al. ArrayMagic: two-colour cDNA microarray quality control and preprocessing Bioinformatics 2005;21:554-556.[Abstract/Free Full Text] 19. Huber W, von Heydebreck A, Sültmann H, et al. Variance stabilization applied to microarray data calibration and to the quantification of differential expression Bioinformatics 2002;18(Suppl 1):S96-S104.[Abstract] 20. Irizarry RA, Hobbs B, Collin F, et al. Exploration, normalization, and summaries of high density oligonucleotide array probe level data Biostatistics 2003;4:249-264.[Abstract] 21. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response Proc Natl Acad Sci U S A 2001;98:5116-5121.[Abstract/Free Full Text] 22. Bussey KJ, Kane D, Sunshine M, et al. MatchMiner: a tool for batch navigation among gene and gene product identifiers Genome Biol 2003;4:R27.[CrossRef][Medline] 23. Al Shahrour F, Diaz-Uriarte R, Dopazo J. FatiGO: a web tool for finding significant associations of Gene Ontology terms with groups of genes Bioinformatics 2004;20:578-580.[Abstract/Free Full Text] 24. 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 2002;99:6567-6572.[Abstract/Free Full Text] 25. Ruschhaupt M, Huber W, Poustka A, et al. A Compendium to ensure computational reproducibility in high-dimensional classification tasks Stat Appl Genet Mol Biol 2004;3:1. 26. Kääb S, Barth AS, Margerie D, et al. Global gene expression in human myocardium-oligonucleotide microarray analysis of regional diversity and transcriptional regulation in heart failure J Mol Med 2004;82:308-316.[CrossRef][Web of Science][Medline] 27. Damas JK, Eiken HG, Oie E, et al. Myocardial expression of CC- and CXC-chemokines and their receptors in human end-stage heart failure Cardiovasc Res 2000;47:778-787.[CrossRef][Web of Science][Medline] 28. Chien KR. Genomic circuits and the integrative biology of cardiac diseases Nature 2000;407:227-232.[CrossRef][Medline] 29. Goetze JP, Gore A, Moller CH, et al. Acute myocardial hypoxia increases BNP gene expression FASEB J 2004;18:1928-1930.[Abstract/Free Full Text] 30. Wiese S, Breyer T, Dragu A, et al. Gene expression of brain natriuretic peptide in isolated atrial and ventricular human myocardium: influence of angiotensin II and diastolic fiber length Circulation 2000;102:3074-3079.[Abstract/Free Full Text] 31. Senbonmatsu T, Saito T, Landon EJ, et al. A novel angiotensin II type 2 receptor signaling pathway: possible role in cardiac hypertrophy EMBO J 2003;22:6471-6482.[CrossRef][Web of Science][Medline] 32. Ruperez M, Lorenzo O, Blanco-Colio LM, et al. Connective tissue growth factor is a mediator of angiotensin II-induced fibrosis Circulation 2003;108:1499-1505.[Abstract/Free Full Text] 33. Omura T, Yoshiyama M, Kim S, et al. Involvement of apoptosis signal-regulating kinase-1 on angiotensin II-induced monocyte chemoattractant protein-1 expression Arterioscler Thromb Vasc Biol 2004;24:270-275.[Abstract/Free Full Text] 34. Schumann H, Holtz J, Zerkowski HR, et al. Expression of secreted frizzled related proteins 3 and 4 in human ventricular myocardium correlates with apoptosis related gene expression Cardiovasc Res 2000;45:720-728.[Abstract/Free Full Text] 35. van Gijn ME, Daemen MJ, Smits JF, et al. The Wnt-frizzled cascade in cardiovascular disease Cardiovasc Res 2002;55:16-24.[Free Full Text] 36. Liu HR, Zhao RR, Jiao XY, et al. Relationship of myocardial remodeling to the genesis of serum autoantibodies to cardiac beta(1)-adrenoceptors and muscarinic type 2 acetylcholine receptors in rats J Am Coll Cardiol 2002;39:1866-1873.[Abstract/Free Full Text] 37. Torre-Amione G. Immune activation in chronic heart failure Am J Cardiol 2005;95:3C-8C.[Web of Science][Medline] 38. Yamamoto T, Eckes B, Mauch C, et al. Monocyte chemoattractant protein-1 enhances gene expression and synthesis of matrix metalloproteinase-1 in human fibroblasts by an autocrine IL-1 alpha loop J Immunol 2000;164:6174-6179.[Abstract/Free Full Text]
Related Article
-
Microarrays Coming of Age in Cardiovascular Medicine: Standards, Predictions, and Biology
- Geoffrey S. Ginsburg, David Seo, and Camille Frazier
J. Am. Coll. Cardiol. 2006 48: 1618-1620.
[Full Text]
[PDF]
This article has been cited by other articles:
|
|
|
|
 
Y.-J. Wei, C.-J. Cui, Y.-X. Huang, X.-L. Zhang, H. Zhang, and S.-S. Hu
Upregulated expression of cardiac ankyrin repeat protein in human failing hearts due to arrhythmogenic right ventricular cardiomyopathy
Eur J Heart Fail,
June 1, 2009;
11(6):
559 - 566.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
U. D. Akavia and D. Benayahu
Meta-analysis and profiling of cardiac expression modules
Physiol Genomics,
November 12, 2008;
35(3):
305 - 315.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
Z. Gao, A. S. Barth, D. DiSilvestre, F. G. Akar, Y. Tian, A. Tanskanen, D. A. Kass, R. L. Winslow, and G. F. Tomaselli
Key pathways associated with heart failure development revealed by gene networks correlated with cardiac remodeling
Physiol Genomics,
November 12, 2008;
35(3):
222 - 230.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
R. Kuner, A. S. Barth, M. Ruschhaupt, A. Buness, L. Zwermann, E. Kreuzer, G. Steinbeck, A. Poustka, H. Sultmann, and M. Nabauer
Genomic analysis reveals poor separation of human cardiomyopathies of ischemic and nonischemic etiologies
Physiol Genomics,
June 1, 2008;
34(1):
88 - 94.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
G. E. Haddad, L. J. Saunders, S. D. Crosby, M. Carles, F. del Monte, K. King, M. R. Bristow, F. G. Spinale, T. E. Macgillivray, M. J. Semigran, et al.
Human cardiac-specific cDNA array for idiopathic dilated cardiomyopathy: sex-related differences
Physiol Genomics,
April 1, 2008;
33(2):
267 - 277.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
A. O. Gramolini, T. Kislinger, R. Alikhani-Koopaei, V. Fong, N. J. Thompson, R. Isserlin, P. Sharma, G. Y. Oudit, M. G. Trivieri, A. Fagan, et al.
Comparative Proteomics Profiling of a Phospholamban Mutant Mouse Model of Dilated Cardiomyopathy Reveals Progressive Intracellular Stress Responses
Mol. Cell. Proteomics,
March 1, 2008;
7(3):
519 - 533.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
G. S. Ginsburg, D. Seo, and C. Frazier
Microarrays Coming of Age in Cardiovascular Medicine: Standards, Predictions, and Biology
J. Am. Coll. Cardiol.,
October 17, 2006;
48(8):
1618 - 1620.
[Full Text]
[PDF]
|
|
|
|
Top
Abstract
Methods