Gene Expression Microarray Studies Research Articles

Abstract A24: Evaluation and validation of biomarkers for prostate cancer early diagnosis

Abstract Prostate cancer (PCa) remains the most frequently diagnosed malignancy in men and although effective treatment is possible for early stages, advanced PCa is largely incurable. PSA, the only relevant marker for this disease, has a low predictive value leading to unnecessary biopsies and associated complications. Thus, identification of better biomarkers able to accurately and reliably diagnose PCa at early stages is an important goal in prostate cancer research. Currently a number of potential biomarkers are being investigated as predictors of PCa diagnosis. Using quantitative real time PCR (qRT-PCR) we have analyzed in the PCa cell lines PWR-1E, LNCaP and PC3 the expression of a set of potential biomarkers previously identified by us in a microarray gene expression study. So far, we have found in these cell lines several differentially expressed genes, some of which encode extracellular matrix proteoglycans, cell adhesion molecules, genes related to motility, growth factor receptors, and tumor suppressor genes. We are in the process of evaluating these differentially expressed genes in cDNA panels of PCa at different disease stages and in prostate needle biopsy specimens taken from patients at diagnosis before any treatment have been established. The next step will be the validation, at the protein level in serum samples from a cohort of patients, of those genes showing an association to prostate cancer disease stages. A prospective clinical study will be necessary to confirm the clinical relevance of validated genes. Project code: 110745921483. Supported by Colciencias Grant # RC No. 462-2008 awarded to Niradiz Reyes. Citation Information: Cancer Prev Res 2011;4(10 Suppl):A24.

A Growth Curve Model with Fractional Polynomials for Analysing Incomplete Time-Course Data in Microarray Gene Expression Studies

Identifying the various gene expression response patterns is a challenging issue in expression microarray time-course experiments. Due to heterogeneity in the regulatory reaction among thousands of genes tested, it is impossible to manually characterize a parametric form for each of the time-course pattern in a gene by gene manner. We introduce a growth curve model with fractional polynomials to automatically capture the various time-dependent expression patterns and meanwhile efficiently handle missing values due to incomplete observations. For each gene, our procedure compares the performances among fractional polynomial models with power terms from a set of fixed values that offer a wide range of curve shapes and suggests a best fitting model. After a limited simulation study, the model has been applied to our human in vivo irritated epidermis data with missing observations to investigate time-dependent transcriptional responses to a chemical irritant. Our method was able to identify the various nonlinear time-course expression trajectories. The integration of growth curves with fractional polynomials provides a flexible way to model different time-course patterns together with model selection and significant gene identification strategies that can be applied in microarray-based time-course gene expression experiments with missing observations.

Synergistic enhancement of cellular immune responses by the novel Toll receptor 7/8 agonist 3M-007 and interferon-γ: implications for therapy of cutaneous T-cell lymphoma

Cutaneous T-cell lymphoma (CTCL) is responsive at all stages to immunotherapy. We determined whether a novel agonist for Toll-like receptor (TLR) 7/8 (3M-007) combined with either interferon-γ (IFN-γ) or interleukin-15 (IL-15) would enhance patients’ immune responses in vitro. Our data demonstrate that IFN-γ or IL-15 in combination with 007 significantly increases patients’ natural killer (NK) cytolytic activity against CTCL tumor cell lines and synergistically induces dendritic cell cytokines, compared to 007 alone. Microarray studies of gene expression of patients’ peripheral blood mononuclear cells (PBMCs) primed with IFN-γ followed by stimulation with 007 identified significant up-regulation of the expression of IL-12 p35 (α-chain), IL-12 p40 (β-chain), and nine IFN-α genes. Importantly, the underlying mechanism of increased levels of IFN-α and IL-12 from combined treatment appears to involve IFN regulatory factor 8 (IRF-8). These results further support our hypothesis that combinations of biological modifiers activating different arms of the immune system may provide significant therapeutic benefits for patients with advanced CTCL.

Melanoma Prognosis: A REMARK-Based Systematic Review and Bioinformatic Analysis of Immunohistochemical and Gene Microarray Studies

Despite intensive research efforts, within-stage survival rates for melanoma vary widely. Pursuit of molecular biomarkers with improved prognostic significance over clinicohistological measures has produced extensive literature. Reviews have synthesized these data, but none have systematically partitioned high-quality studies from the remainder across different molecular methods nor examined system properties of that output. Databases were searched for studies analyzing protein expression by immunohistochemistry (n = 617, extending the only systematic review to date by 102 studies) or for gene expression microarray studies (n = 45) in melanoma in relation to outcome. REMARK-derived criteria were applied to identify high-quality studies. Biomarkers and pathways were functionally assessed by using gene ontology software. Most manuscripts did not meet REMARK-based criteria, an ongoing trend that can impede translational research. Across REMARK-compliant literature, 41 proteins were significantly associated with outcome. Multimarker tests consistently emerged among the most promising potential biomarkers, indicating a need to continue assessing candidates in that composite setting. Twenty-one canonical pathways were populated by outcome-related proteins but not by those that failed to show such an association; we propose that this set of pathways warrants closer investigation to understand drivers of poor outcome in melanoma. Two-gene expression microarray studies met REMARK-based criteria reflecting a genuine paucity of literature in the area. The 254 outcome-related genes were examined for correspondences with the systematically identified protein signature. This analysis highlighted proliferating cell nuclear antigen and survivin as priorities for further examination as biomarkers in melanoma prognosis, and illustrated ongoing need to integrate alternative approaches to biomarker discovery in melanoma translational research.

Targeting Androgen Receptor in Estrogen Receptor-Negative Breast Cancer

Endocrine therapies for breast cancer that target the estrogen receptor (ER) are ineffective in the 25%-30% of cases that are ER negative (ER-). Androgen receptor (AR) is expressed in 60%-70% of breast tumors, independent of ER status. How androgens and AR regulate breast cancer growth remains largely unknown. We find that AR is enriched in ER- breast tumors that overexpress HER2. Through analysis of the AR cistrome and androgen-regulated gene expression in ER-/HER2+ breast cancers we find that AR mediates ligand-dependent activation of Wnt and HER2 signaling pathways through direct transcriptional induction of WNT7B and HER3. Specific targeting of AR, Wnt or HER2 signaling impairs androgen-stimulated tumor cell growth suggesting potential therapeutic approaches for ER-/HER2+ breast cancers.

Up-Regulation of the Cardiac Lipid Metabolism at the Onset of Heart Failure

Chronic pressure overload and atherosclerosis are primary etiologic factors for cardiac hypertrophy and failure. However, mechanisms underlying the transition from hypertrophy to heart failure are incompletely understood. We analyzed the development of heart failure in mice with chronic pressure overload induced by aortic constriction and compared the results with aged apolipoprotein E-deficient mice suffering from advanced atherosclerosis. We combined cardiac function analysis by echocardiography and invasive hemodynamics with a comprehensive microarray gene expression study (GSE25765-8). The microarray data showed that the onset of heart failure induced by pressure overload or advanced atherosclerosis was accompanied by a strong up-regulation of key lipid metabolizing enzymes involved in fat synthesis, storage and oxidation. Cardiac lipid overload may be involved in the progression of heart failure by enhancing cardiomyocyte death. Up-regulation of the cardiac lipid metabolism was related to oxygen and ATP depletion of failing hearts because anti-ischemic treatment with ranolazine normalized the cardiac lipid metabolism and improved cardiac function. Vice versa, inhibition of cellular respiration and ATP generation by mild thiol-blocking with cystamine triggered the cardiac lipid metabolism and caused signs of heart failure. Cardiac tissue specimens of patients with heart failure also showed high protein levels of key fat metabolizing enzymes as well as lipid accumulation. Taken together, our data strongly indicate that up-regulation of the cardiac lipid metabolism and myocardial lipid overload are underlying the development of heart failure.

Variations in aggrecan localization and gene expression patterns characterize increasing stages of human intervertebral disk degeneration

During disk degeneration, annulus dehydration and matrix fraying culminate in the formation of tears through which nucleus and annulus disk material may rupture, causing radicular pain. Annular tears are present in more than half of the patients in early adulthood and are almost always present in the elderly. Aggrecan, which provides the disk with a shock absorber function under loading, is a key disk extracellular matrix (ECM) component. The objective of the present study was to assess the immunolocalization of aggrecan in the annulus, and to assess molecular gene expression patterns in the annulus ECM utilizing microarray analysis. Immunohistochemistry was performed on 45 specimens using an anti-human aggrecan antibody. Affymetrix microarray gene expression studies used the extracellular matrix ontology approach to evaluate an additional 6 grade I–II, 9 grade III, and 4 grade IV disks. Grade III/IV disks were compared to healthier grade I/II disks. Healthy and less degenerated disks showed a general uniform aggrecan immunolocalization; more degenerated disks contained regions with little or no identifiable aggrecan localization. In degenerated disks, molecular studies showed a significant downregulation of aggrecan, ADAMTS-like 3, and ADAMTS10. Collagen types III and VIII, fibronectin, decorin, connective tissue growth factor, TIMP-3, latent TGF-β binding protein 2 and TGF-β1 were significantly upregulated with fold changes ranging from 2.4 to 9.8. Findings here help us better understand changes in the immunohistochemical distribution of a key proteoglycan during disk aging. Such information may have application as we work towards biologic therapies to improve the aging/degenerating disk matrix.

An Empirical Bayes' Approach to Joint Analysis of Multiple Microarray Gene Expression Studies

With the prevalence of gene expression studies and the relatively low reproducibility caused by insufficient sample sizes, it is natural to consider joint analysis that could combine data from different experiments effectively to achieve improved accuracy. We present in this article a model-based approach for better identification of differentially expressed genes by incorporating data from different studies. The model can accommodate in a seamless fashion a wide range of studies including those performed at different platforms by fitting each data with different set of parameters, and/or under different but overlapping biological conditions. Model-based inferences can be done in an empirical Bayes' fashion. Because of the information sharing among studies, the joint analysis dramatically improves inferences based on individual analysis. Simulation studies and real data examples are presented to demonstrate the effectiveness of the proposed approach under a variety of complications that often arise in practice.

Hedgehogs, Flies, Wnts and MYCs: The Time Has Come for Many Things in Medulloblastoma

Medulloblastoma epitomizes the war against childhood cancer. Although this small round blue-cell tumor accounts for only 20% of all childhood brain tumors, or just fewer than 600 new cases each year in the United States, medulloblastoma has nevertheless become a paradigm for biologic study, clinical investigation, and assessment of late effects. Craniospinal irradiation historically became the mainstay of therapy for this WHO grade 4 neoplasm because of its propensity to disseminate throughout the neuraxis via the subarachnoid space. By the 1970s, 36 Gy of craniospinal radiotherapy resulted in 5-year survival of greater than 50% for better-risk patients. With cooperative group trials in the 1980s, adjuvant chemotherapy was introduced to medulloblastoma treatment protocols (SIOP I [International Society of Paediatric Oncology], CCG 942 [Children’s Cancer Group], and POG 7909 [Pediatric Oncology Group]), leading to improved survival and subsequent reduction of radiotherapy dosage for averagerisk patients in the 1990s and early part of the last decade. A decrease in craniospinal radiotherapy from 36 Gy to 23.4 Gy resulted in better neurocognitive and physical outcomes for average-risk patients, highlighting the importance of developing appropriately riskadapted therapy. Although prognosis has improved over many years, a significant fraction of children still do not respond to standard therapy. Perhaps equally disturbing, survivors are left with a multitude of sequelae attributable to chemotherapy and radiation therapy, including cognitive impairment, hearing loss, neuropathies, seizures, endocrinopathies, strokes, and second malignancies. Thus, there is an overwhelming need to refine strategies that will minimize unnecessarily aggressive therapy for patients with low-risk disease and maximize efficacious treatment for patients with high-risk disease. Such refinement can come only with better disease or risk classification. In an era of genomic profiling, risk stratification for medulloblastoma disappointingly remains strictly clinical, based solely on age, resection extent, Chang metastasis staging, and more recently anaplasia. Since the mid 1990s, a two-tiered risk stratification system has been employed for the purposes of North American cooperative group trials. Patients older than 3 years with less than 1.5 cm of residual tumor and nonmetastatic disease are considered average risk, whereas those younger than 3 years with more than 1.5 cm of residual tumor, evidence of tumor dissemination, or diffuse anaplasia are considered high risk. Multiple histopathologic, cytogenetic, and molecular factors have been found to influence prognosis, but to date, none have been incorporated into the risk-stratification system. Histopathologically, the desmoplastic variant has long been recognized as connoting better prognosis, whereas the large-cell anaplastic variant more recently has been found to be associated with reduced survival. Moreover, isochromosome 17q, loss of heterozygosity at chromosome 17p, and high-level chromosomal gain at 8q24 (locus of c-MYC) have all been associated with large-cell anaplastic variant medulloblastoma and poor prognosis. Monosomy 6 has been associated with favorable prognosis. In addition, a number of molecular markers have been identified over the past decade and have laid the groundwork for a risk-stratification system with a biologic basis. Favorable prognosis has been described for children with high TrkC expression or nucleopositivity of -catenin, an indicator of activated Wnt signaling—a pathway now known to affect embryogenesis and cancer. Poor prognosis is indicated by overexpression of calbindin-D, elevated ErbB2 expression, and elevated expression of c-MYC, a member of the MYC family of transcription factors that stimulates cell proliferation. In recent years, large-scale genomic and gene expression profile studies have begun to parse the molecular heterogeneity of medulloblastomas into clear subtypes with associated risk stratification. Initial gene expression microarray studies used a supervised approach, grouping samples on the basis of histology, metastatic status, and survival to identify differentially expressed genes in these clinical and histologic groups. More recently, unsupervised microarray studies in medulloblastoma by Kool et al (62 samples) and Thompson et al (46 samples) each identified up to five subgroups of medulloblastoma, including a subgroup characterized by activation of the Wnt signaling pathway, associated with excellent prognosis, and a subgroup characterized by activation of the Hedgehog (Hh) signaling pathway occurring primarily in infants and adults and associated with desmoplastic histology. The Hh signaling pathway is involved in organogenesis, limb formation, brain patterning, and multiple forms of cancer. There are several Hh family ligands, the best known of which is Sonic Hedgehog (SHh) for its multiple developmental roles, including in cerebellar development. The study by Kool et al also identified a molecular subtype associated with a high rate of metastatic disease, anaplastic/ large-cell histology, and poor prognosis. In this issue of Journal of Clinical Oncology, four separate studies dovetail with previous work to crystallize a molecular classification of medulloblastoma and codify a new risk-stratification model on the basis of the molecular biology of these disease subtypes. Ellison et al JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L S VOLUME 29 NUMBER 11 APRIL 1

Gene set enrichment analysis of microarray data from Pimephales promelas (Rafinesque), a non-mammalian model organism

BackgroundMethods for gene-class testing, such as Gene Set Enrichment Analysis (GSEA), incorporate biological knowledge into the analysis and interpretation of microarray data by comparing gene expression patterns to pathways, systems and emergent phenotypes. However, to use GSEA to its full capability with non-mammalian model organisms, a microarray platform must be annotated with human gene symbols. Doing so enables the ability to relate a model organism's gene expression, in response to a given treatment, to potential human health consequences of that treatment. We enhanced the annotation of a microarray platform from a non-mammalian model organism, and then used the GSEA approach in a reanalysis of a study examining the biological significance of acute and chronic methylmercury exposure on liver tissue of fathead minnow (Pimephales promelas). Using GSEA, we tested the hypothesis that fathead livers, in response to methylmercury exposure, would exhibit gene expression patterns similar to diseased human livers.ResultsWe describe an enhanced annotation of the fathead minnow microarray platform with human gene symbols. This resource is now compatible with the GSEA approach for gene-class testing. We confirmed that GSEA, using this enhanced microarray platform, is able to recover results consistent with a previous analysis of fathead minnow exposure to methylmercury using standard analytical approaches. Using GSEA to compare fathead gene expression profiles to human phenotypes, we also found that fathead methylmercury-treated livers exhibited expression profiles that are homologous to human systems & pathways and results in damage that is similar to those of human liver damage associated with hepatocellular carcinoma and hepatitis B.ConclusionsThis study describes a powerful resource for enabling the use of non-mammalian model organisms in the study of human health significance. Results of microarray gene expression studies involving fathead minnow, typically used for aquatic ecological toxicology studies, can now be used to generate hypotheses regarding consequences of contaminants and other stressors on humans. The same approach can be used with other model organisms with microarray platforms annotated in a similar manner.

Adaptive Elastic-Net Sparse Principal Component Analysis for Pathway Association Testing

Pathway or gene set analysis has become an increasingly popular approach for analyzing high-throughput biological experiments such as microarray gene expression studies. The purpose of pathway analysis is to identify differentially expressed pathways associated with outcomes. Important challenges in pathway analysis are selecting a subset of genes contributing most to association with clinical phenotypes and conducting statistical tests of association for the pathways efficiently. We propose a two-stage analysis strategy: (1) extract latent variables representing activities within each pathway using a dimension reduction approach based on adaptive elastic-net sparse principal component analysis; (2) integrate the latent variables with the regression modeling framework to analyze studies with different types of outcomes such as binary, continuous or survival outcomes. Our proposed approach is computationally efficient. For each pathway, because the latent variables are estimated in an unsupervised fashion without using disease outcome information, in the sample label permutation testing procedure, the latent variables only need to be calculated once rather than for each permutation resample. Using both simulated and real datasets, we show our approach performed favorably when compared with five other currently available pathway testing methods.

Gene Expression Profiling with Principal Component Analysis Depicts the Biological Continuum From Essential Thrombocythemia Over Polycythemia Vera to Myelofibrosis

AbstractAbstract 4115 Introduction:The classical Philadelphia-negative chronic myeloproliferative neoplasms (CMPNs) encompass polycythemia vera (PV), essential thrombocythemia (ET), and primary myelofibrosis (PMF). After several years, a proportion of patients with PV may develop increasing splenomegaly and bone marrow fibrosis to terminate in postpolycythemic myelofibrosis, which then has a very poor prognosis with median survival of only a few years. A high proportion of the ET-population has been shown to have prefibrotic PMF or a “PV-like” phenotype, which after a variable time may terminate in myelofibrosis with myeloid metaplasia (MMM). Although the concept of prefibrotic myelofibrosis as a separate disease entity to be distinguished from ET by distinct histopathological features has been questioned, it seems logical that a continuum exists from “early” myelofibrosis with no or minimal bone marrow fibrosis to the advanced stage of myelofibrosis with myeloid metaplasia. Accordingly, a new concept of these diseases as a biological continuum from ET over PV to PMF has emerged. Indeed, this biological continuum reflects the various clinical stages as also described for chronic myelogenous leukaemia (CML) – a chronic stable phase (ET-PV), an accelerated phase (transitional phase), and a terminal myelofibrosis or blast phase of the disease (PMF-BP).It is evident that the JAK2 V617F mutational load is an important determinant of phenotype, but other genetic and epigenetic events contribute to the phenotypic presentation. Several gene expression studies have identified genes which might be of pathogenetic relevance for the development of CMPNs. However, no study has applied an unsupervised method as Principal Component Analysis (PCA) to discover unknown trends in the data. We have carried out gene expression profiling with PCA to ascertain if this analysis might unravel distinct grouping of disease entities supporting the biological continuum from early ET over PV to the advanced myelofibrosis phase. Patients and Methods:Using HG-U133 Plus 2.0 microarray from Affymetrix, recognizing 54675 probe sets (38500 genes), gene expression microarray studies have been performed on control subjects (n=21) and patients with ET (n =19), PV (n=41), and PMF (n=9). An unsupervised statistical method, PCA, which is an exploratory tool used to uncover unknown trends in the data based on gene expression profiles, was applied to the data. Results:The results from the PCA confirm the hypothesis of a biological continuum from control subjects over ET to PV and finally PMF (figure 1). The figure also reveals that patients with PV in a transitional stage (PVtrans) with pancytosis, huge splenomegaly, and bone marrow fibrosis have a gene profile between PV and PMF. Furthermore, single gene analysis revealed that several genes are highly dysregulated in patients with ET, PV, and PMF compared to control subjects even after correcting for multiple testing. Discussion and Conclusions:Using global gene expression profiling with PCA in patients with different CMPNs (ET, PV, PMF), we have yielded support to the concept of a biological continuum from early ET to the advanced myelofibrosis stage. Studies are in progress to identify single genes or clusters of genes accounting for the biological continuum and accordingly the progression from ET over PV to myelofibrosis. Disclosures:No relevant conflicts of interest to declare.

High Expression of Carcinoembryonic Antigen-Related Cell Adhesion Molecule(CEACAM) 6 In Primary Myelofibrosis

AbstractAbstract 4116 Introduction:Carcinoembryonic antigen-related cell adhesion molecule (CEACAM) 6, also known as non-specific cross-reacting antigen (NCA) or CD66c is a glycosylphosphoinositol (GPI)-linked cell surface protein and a member of the CEA family of proteins. Structurally, this protein shares close homology with CEACAM1, CEACAM7 and CEACAM8. Functionally, CEACAM6 has been implicated in cell adhesion, cellular invasiveness and metastatic behaviour of tumour cells. Expression of CEACAM6 protein has been found in a variety of normal human tissues, including myeloid cells. A key pathophysiological feature of primary myelofibrosis (PMF) is changes in the bone marrow micromilieu, progressive accumulation of connective tissue, pronounced neovascularisation and altered stromal cell adhesion. Consequently, CD34+ cells escape the bone marrow and seed extramedullarily. Being involved in cell adhesion, cellular invasiveness, angiogenesis, and inflammation – all key processes in the pathophysiology of PMF – we hypothesized that CEACAM6 might play an important role in these processes in patients with myelofibrosis. We have assessed gene expression of several CEA genes in patients with PMF and related neoplasms in order to elucidate the significance of CEACAM6 and other members of the CEA family of proteins in these disorders. Patients and Methods:Gene expression microarray studies have been performed on whole blood from control subjects (n=21) and patients with ET (n =19), PV (n=41), and PMF (n=9). Gene expression profiles were generated using Affymetrix HG-U133 2.0 Plus microarrays recognizing 54.675 probe sets (38.500 genes). Total RNA was purified from whole-blood and amplified to biotin-labeled aRNA and hybridized to microarray chips. Results:20.439, 25.307, 17.417, and 25.421 probe sets were identified to be differentially expressed between controls and patients with ET, PV, PMF, and CPMNs as a whole, respectively (false discovery rate (FDR) adjusted p values < 0.05). Several CEACAM-genes were significantly deregulated. In PMF patients, the CEACAM genes 1, 6 and 8 were significantly upregulated with the highest upregulation of CEACAM6 and CEACAM8 (fold change (FC) 12.5 and 14.0, respectively and FDR adjusted p values 7.71 × 10-7 and 1.48 × 10-5, respectively). Only the CEACAM21 gene was significantly downregulated (FC -1.3 and FDR adjusted p-value 4.14 × 10-7) whereas the other CEACAM-genes tested (3, 4, 5, 7, 19) displayed no significant changes as compared to controls. In ET patients, the CEACAM genes 3, 6, and 7 were significantly upregulated (FC 1.2, 1.8, and 1.1, respectively) and FDR adjusted p values < 0.03). The CEACAM21 gene was significantly downregulated (FC -1.3 and FDR adjusted p-value 0.0004). In PV patients, the CEACAM genes 1, 3, 6 and 7 were significantly upregulated (FC 1.7, 1.2, 1.7, and 1.1, respectively) and FDR adjusted p values 0.0002, 0.0009, 0.0002, and 0.03, respectively. The CEACAM21 gene was significantly downregulated (FC -1.3 and FDR adjusted p-value 4.14×10-5) All other CEACAM-genes showed no significant changes in either ET or PV as compared to controls.When comparing controls with non-PMF-patients, a significant upregulation of the CEACAM genes 1, 3, 6, and 7 were recorded in non-PMF patients (FC 1.5, 1.2, 1.7, and 1.1, respectively; FDR adjusted p values 0.001, 0.0002, 0.002, and 0.02, respectively). The CEACAM19 and CEACAM21 genes were significantly downregulated (FC - 1.1 and -1.4, respectively; adjusted p-values 0.008 and 3.46 × 10-8). Discussion and Conclusions:Using global gene expression profiling, we have found a pronounced deregulation of CEACAM genes, involving highly significant upregulation of the CEACAM genes 6 and 8 in PMF (FCs 12.0 and 14.0, respectively). Upregulation of CEACAM6 was seen in both ET, PV and PMF by far the highest levels being recorded in PMF-patients. Of note, significant upregulation of CEACAM8 (FC 14) was only seen in patients with myelofibrosis. The elevated expression of CEACAMs genes in ET, PV, and PMF may solely reflect neutrophil activation being most exaggerated in patients with PMF in whom the highest CEACAM6 and 8 expression patterns were recorded. Alternatively, the highly elevated gene expression of CEACAM6 and 8 in PMF are molecular markers of clonal expansion and myelofibrotic transformation, implying enhanced proteolytic activity and egress of CD34+ cells into the circulation. Disclosures:No relevant conflicts of interest to declare.

Increased Gene Expression of Histone Deacetylases In Patients with Philadelphia-Negative Chronic Myeloproliferative Neoplasms

AbstractAbstract 4119 Introduction:Several new treatment strategies within the Philadelphia-negative chronic myeloproliferative neoplasms CMPNs are beeing explored, among these, agents belonging to the class of HDAC-inhibitors, including givinostat (ITF2357) and vorinostat (suberoylanilide hydroxamic acid (SAHA)). These agents are inhibitors of class I and II HDAC enzymes, promoting cell-cycle arrest and apoptosis of cancer cells. Recently, enhanced histone deacetylase (HDAC) enzyme activity has been found in CD34+ cells from patients with primary myelofibrosis (PMF), enzyme activity levels highly exceeding those recorded in essential thrombocythemia (ET) and polycythemia vera (PV). The raised levels correlated to the degree of splenomegaly, suggesting that HDAC might be recruited as ET or PV progresses into myelofibrosis or PMF progresses into a more advanced stage. In order to further describe HDACs in CMPNs, we have assessed gene expression of several HDACs in a larger cohort of patients with ET, PV and PMF. Patients and Methods:Gene expression microarray studies have been performed on control subjects (n=21) and patients with ET (n =19), PV (n=41), and PMF (n=9). Most patients were studied on cytoreductive therapy, which for the large majority included hydroxyurea. Gene expression profiles were generated using Affymetrix HG-U133 2.0 Plus microarrays recognizing 54675 probe sets (38.500 genes). Total RNA was purified from whole blood and amplified to biotin-labeled aRNA and hybridized to microarray chips. Results:We identified 20439, 25307, 17417, and 25421 probe sets which were differentially expressed between controls and patients with ET, PV, PMF, and CPMNs as a whole, respectively (false discovery rate (FDR) adjusted p values < 0.05). Several HDAC-genes were significantly deregulated. In ET patients upregulated genes included HDAC5, HDAC9 and downregulated genes HDAC1, HDAC4, HDAC7. In PV-patients HDAC4, HDAC5, HDAC6, HDAC9, and HDAC11 genes were upregulated, and HDAC1, HDAC7, HDAC9, and HDAC11 genes were downregulated. In PMF -patients HDAC4, HDAC6, HDAC9, and HDAC11 genes were upregulated, and HDAC1 and HDAC7 were downregulated. Compared to controls the CMPN-patients as a group exhibited upregulation of HDAC4, HDAC5, HDAC6, HDAC9, and HDAC11 genes. The HDAC genes 9 and 11 were significantly upregulated in both ET, PV, PMF, and CMPNs as a whole, the highest values being recorded in patients with ET, PMF, and CMPNs as compared to controls. In regard to HDAC9 gene expression, the fold changes (FC) were 1.3, 1.2, 1.3, and 1.3 for ET, PV, PMF, and CMPNs, respectively; for HDAC11 the highest values were recorded in patients with ET with the following FCs 1.2, 1.1, 1.1, 1.1 for ET, PV, PMF, and CMPNs, respectively; FDR adjusted p values < 0.05). Within patients, the HDAC6-gene was also differentially expressed with the highest levels being recorded in patients with PMF (FC 1.2, FDR adjusted p values < 0.01). When comparing non-PMF-patients with PMF-patients, a significant upregulation of the HDAC2-gene was found in PMF patients (FC 1.5, FDR adjusted p-value=0.007), whereas the HDAC7-gene was significantly downregulated (FC -1.3, FDR adjusted p-value=0,001. Discussion and Conclusions:Using global gene expression profiling of whole-blood from patients with CMPNs, we have found a pronounced deregulation of HDAC-genes, involving significant upregulation of the HDAC-genes 9 and 11 with the highest expression levels being found in patients with ET, in PMF (HDAC9) and in CMPNs both HDAC9 and HDAC11. Furthermore, we have identified that the HDAC-6 gene is progressively expressed in patients with ET, PV, and PMF reflecting a steady accumulation of abnormally expressed HDAC-6 during disease evolution. Our results lend further support to HDACs as important epigenetic targets in the future treatment of patients with CMPNs. Since the highest expression levels of HDAC-genes was recorded in ET (HDACs 9 and 11), in PMF (HDAC9) and in the whole CMPN-group both HDACs 9 and 11, their downregulation by HDAC-inhibitors might be associated with decreased disease activity including reduction of splenomegaly in MF. Preliminary data indicate that enlarged spleens are diminished in subsets of patients during treatment with HDAC-inhibitors. Studies are in progress on the impact of vorinostat on global gene expression profiling, including HDAC- gene expression subclasses in patients with CMPNs. Disclosures:No relevant conflicts of interest to declare.

Integrative Genome-Wide DNA Methylation and Gene Expression Analysis Reveals Biological and Clinical Insights In Adult Acute Lymphoblastic Leukemia

AbstractAbstract 852Adult acute lymphoblastic leukemia (ALL) is an aggressive disease with <30% long-term survival. This relatively poor outcome can be explained in part by an increased frequency of high-risk molecular subtypes compared to childhood ALL, such as BCR/ABL (20%-40% in adults vs 2%-5% in children). We hypothesized that aberrant epigenetic gene regulation contributes to the pathogenesis and clinical features of adult ALL. We therefore performed genome-wide DNA methylation and gene expression microarray studies of 215 adult patients with B-lineage ALL enrolled in the ECOG E2993 phase III trial. Patients had a median follow-up 4.75 years (3.5 months to 13 years) and median age 39 years (17 to 63 years). BCR/ABL(+) cases (n=83) had a worse overall survival (OS, p=0.08) than BCR/ABL(-) cases (n=132). The smaller difference in survival in this series between BCR/ABL(+) and (-) cases is likely due to the use of imatinib in some of these patients. The HELP microarray assay was used to measure DNA methylation at 50,000 CpGs annotated to ∼22,000 RefSeq promoters. The accuracy of HELP was confirmed by extensive quantitative single locus validation studies. Supervised analysis revealed the presence of a markedly aberrant DNA methylation signature (166 genes) in BCR/ABL(+) ALL with the cutoff values of p<0.001 (FDR<0.002) and log fold change>1, and a differential gene expression profile of 416 genes at p<0.001 (FDR<0.004) and log fold change>1. Integrative analysis of expression and DNA methylation indicated that many of these genes are functionally connected within a gene network centered around IL8, which was over expressed and hypomethylated in BCR/ABL(+) cases, along with IL2RA(CD25), CEBPB, ABL1, ID1, IL15, BCL2L13, CD69, NOV, S100A8 and S100A9. KEGG and BioCarta pathway analysis also showed enrichment for IL8 signaling, NF-kB Activation, B Cell Development and Antigen Presentation pathways, suggesting these cytokine networks might play central and distinct roles in BCR/ABL(+) ALL. To identify a core set of functionally relevant genes, we explored the overlap of the DNA methylation and gene expression signatures. The overlap consisted of 13 genes, among which 11 showed inverse correlation, including CD200, GAB1, HLA-DQA1, HLA-DQB1, IL2RA (CD25), LST1, LTB, NOV, ROB04, S100A9, and CD38. Consistent with previous ECOG findings, CD25 positivity was a more dominant predictor than BCR/ABL, and could further stratify BCR/ABL(+) patients into a favorable (CD25-) and a poor (CD25+) outcome group (OS, p=0.07). When comparing CD25(+) and CD25(-) groups among BCR/ABL(+) cases, we found RhoH and UBE2J1 as the top differentially methylated and IL2RA(CD25) as the top differentially expressed genes, and all three genes showed concordant corresponding changes in expression and methylation. The aberrant DNA methylation signature of MLL/AF4 cases was even more dramatic, with 469 identified as differentially methylated (p<0.001 (FDR<0.015), log fold change>1) and 1108 genes differentially expressed (p<0.001 (FDR<0.009), log fold change>1). Integrated analysis of DNA methylation and expression implicated gene pathways centered around TNFA and MYC, respectively. A core set of 44 genes were identified overlapping between the methylation and gene expression signatures, among which 34 showed inverse correlation, including ANXA5, BRE, CAPG, CEBPA, FAIM, FLT3, FUT4, IGFBP7, IL1R2, ITGA7, ITGAE, MAP1A, MAP7, MRPL33, PARP8, RBKS, SLITRK4 and TFR2 with overexpression and hypomethylation, and BTBD3, CCR6, FYN, GAB1, GYPC, HPS4, IL2RA, KCNK3, LCK, LST1, LTB, PRKCH, QPCT, S100A13, SGPL1 and ZAP70 with underexpression and hypermethylation in MLL/AF4(+) ALL. All signatures were independent of B-ALL differentiation stage. Using univariate Cox Hazard Regression model adjusted by age, WBC, CD25 and BCR/ABL status, we furthermore identified 259 expression and 115 methylation markers which were significantly correlated with patient OS risk (p<0.01). Molecularly or immunophenotypically defined poor prognosis adult B- ALLs thus feature specific aberrant DNA methylation profiles with associated inversely correlated gene expression. These gene sets delineate specific biological functions that may contribute to disease phenotype and offer an opportunity for development of targeted therapy. Aberrantly methylated genes in adult B-ALL correlate with clinical risk independent of other clinical or molecular risk factors. Disclosures:No relevant conflicts of interest to declare.

Ranking prognosis markers in cancer genomic studies

In cancer research, high-throughput genomic studies have been extensively conducted, searching for markers associated with cancer diagnosis, prognosis and variation in response to treatment. In this article, we analyze cancer prognosis studies and investigate ranking markers based on their marginal prognosis power. To avoid ambiguity, we focus on microarray gene expression studies where genes are the markers, but note that the methodology and results are applicable to other high-throughput studies. The objectives of this study are 2-fold. First, we investigate ranking markers under three commonly adopted semiparametric models, namely the Cox, accelerated failure time and additive risk models. Data analysis shows that the ranking may vary significantly under different models. Second, we describe a nonparametric concordance measure, which has roots in the time-dependent ROC (receiver operating characteristic) framework and relies on much weaker assumptions than the semiparametric models. In simulation, it is shown that ranking using the concordance measure is not sensitive to model specification whereas ranking under the semiparametric models is. In data analysis, the concordance measure generates rankings significantly different from those under the semiparametric models.

Recent molecular genetic studies and methodological issues in suicide research

Suicide behavior (SB) spans a spectrum ranging from suicidal ideation to suicide attempts and completed suicide. Strong evidence suggests a genetic susceptibility to SB, including familial heritability and common occurrence in twins. This review addresses recent molecular genetic studies in SB that include case-control association, genome gene-expression microarray, and genome-wide association (GWA). This work also reviews epigenetics in SB and pharmacogenetic studies of antidepressant-induced suicide. SB fulfills criteria for a complex genetic phenotype in which environmental factors interact with multiple genes to influence susceptibility. So far, case-control association approaches are still the mainstream in SB genetic studies, although whole genome gene-expression microarray and GWA studies have begun to emerge in recent years. Genetic association studies have suggested several genes (e.g., serotonin transporter, tryptophan hydroxylase 2, and brain-derived neurotrophic factor) related to SB, but not all reports support these findings. The case-control approach while useful is limited by present knowledge of disease pathophysiology. Genome-wide studies of gene expression and genetic variation are not constrained by our limited knowledge. However, the explanatory power and path to clinical translation of risk estimates for common variants reported in genome-wide association studies remain unclear because of the presence of rare and structural genetic variation. As whole genome sequencing becomes increasingly widespread, available genomic information will no longer be the limiting factor in applying genetics to clinical medicine. These approaches provide exciting new avenues to identify new candidate genes for SB genetic studies. The other limitation of genetic association is the lack of a consistent definition of the SB phenotype among studies, an inconsistency that hampers the comparability of the studies and data pooling. In summary, SB involves multiple genes interacting with non-genetic factors. A better understanding of the SB genes by combining whole genome approaches with case-control association studies, may potentially lead to developing effective screening, prevention, and management of SB.

Class prediction for high-dimensional class-imbalanced data

BackgroundThe goal of class prediction studies is to develop rules to accurately predict the class membership of new samples. The rules are derived using the values of the variables available for each subject: the main characteristic of high-dimensional data is that the number of variables greatly exceeds the number of samples. Frequently the classifiers are developed using class-imbalanced data, i.e., data sets where the number of samples in each class is not equal. Standard classification methods used on class-imbalanced data often produce classifiers that do not accurately predict the minority class; the prediction is biased towards the majority class. In this paper we investigate if the high-dimensionality poses additional challenges when dealing with class-imbalanced prediction. We evaluate the performance of six types of classifiers on class-imbalanced data, using simulated data and a publicly available data set from a breast cancer gene-expression microarray study. We also investigate the effectiveness of some strategies that are available to overcome the effect of class imbalance.ResultsOur results show that the evaluated classifiers are highly sensitive to class imbalance and that variable selection introduces an additional bias towards classification into the majority class. Most new samples are assigned to the majority class from the training set, unless the difference between the classes is very large. As a consequence, the class-specific predictive accuracies differ considerably. When the class imbalance is not too severe, down-sizing and asymmetric bagging embedding variable selection work well, while over-sampling does not. Variable normalization can further worsen the performance of the classifiers.ConclusionsOur results show that matching the prevalence of the classes in training and test set does not guarantee good performance of classifiers and that the problems related to classification with class-imbalanced data are exacerbated when dealing with high-dimensional data. Researchers using class-imbalanced data should be careful in assessing the predictive accuracy of the classifiers and, unless the class imbalance is mild, they should always use an appropriate method for dealing with the class imbalance problem.

Effect of Multikinase Inhibitors on Caspase-Independent Cell Death and DNA Damage in HER2-Overexpressing Breast Cancer Cells

The receptor tyrosine kinase, HER2, is overexpressed in approximately 25% of patients with breast cancer and is implicated in the aggressiveness of cancer. Targeting of HER2 signaling with trastuzumab, a monoclonal antibody that inhibits HER2 activity, has demonstrated clinical benefits. We investigated whether the antitumor activity of trastuzumab can be potentiated by dasatinib, a small-molecule tyrosine kinase inhibitor, on breast cancer cell lines that overexpress HER2 (BT474 and SKBR3) or have normal HER2 expression (MCF7 and T47D). Functional, biochemical, and gene expression microarray studies were performed to test the effect of trastuzumab, dasatinib, or a combination of trastuzumab and dasatinib on cell proliferation; HER activation; cell cycle; DNA damage; and apoptosis. The effect of drugs on mice (n = 6 per group) bearing xenograft tumors originating from HER2-overexpressing BT474 cells was assessed, and tumors were evaluated for an effect on volume, HER signaling, and DNA damage. All statistical tests were two-sided. Trastuzumab and dasatinib combination showed a synergistic effect on the proliferation of HER2-overexpressing breast cancer cells (combination index = 0.44, 95% confidence interval = 0.30 to 0.58). The drug combination also induced a stronger inhibitory effect on HER2 activation than the individual drugs, decreased the level of proteins involved in DNA damage response, induced DNA double-strand breaks, cell cycle arrest, and caspase-independent apoptosis. Mice (n = 6 per group) bearing xenograft tumors originating from HER2-overexpressing BT474 cells showed statistically significantly reduced tumor volume on day 28 when treated with the drug combination (control vs trastuzumab and dasatinib combination; mean volume = 2.6 vs 0.5 cm(3), difference = 2.1 cm(3), 95% confidence interval = 0.76 to 3.51 cm(3), P = .01) and total regression of tumors by day 36 with no later relapse. Results showed that HER2 and dasatinib-sensitive tyrosine kinases act in a synergistic manner to safeguard the breast cancer cells from DNA damage. The therapeutic targeting of multikinase inhibition opens new avenues for the treatment of HER2-positive breast cancer patients.

Rank aggregation methods

AbstractThis article provides an overview of rank aggregation methods and algorithms, with an emphasis on modern biological applications. Rank aggregation methods have traditionally been used extensively in marketing and advertisement research, and in applied psychology in general. In recent years, rank aggregation methods have emerged as an important tool for combining information from different Internet search engines or from different omics‐scale biological studies. We discuss three classes of methods, namely distributional based, heuristic, and stochastic search. The original Thurstone's scaling and its extensions represent the first class of methods that are most appropriate for aggregating many short ranked lists. Aggregating results from consumer rankings of products falls into this category of problems. Its application to biological problems is also being explored. On the other hand, heuristic algorithms and stochastic search methods are applicable to the situation of aggregating a small number of long lists, the so‐called ‘high‐level’ meta‐analysis scenario. Combining results from different search engines/criteria and a number of omics‐scale biological applications fall into this category. Heuristic algorithms are deterministic in nature, ranging from simple arithmetic averages of ranks to Markov chains and stationary distributions. Stochastic search algorithms, on the other hand, aim at maximizing a particular criterion such as that following the Kemeny guideline. Several examples will be provided to illustrate, compare, and contrast the methods and algorithms. The examples range from simple and contrive to representing realistic scenarios. In particular, an application to aggregating results from gene expression microarray studies is provided to demonstrate applications of the methods to modern biological problems. Copyright © 2010 John Wiley &amp; Sons, Inc.This article is categorized under: Statistical and Graphical Methods of Data Analysis &gt; Robust Methods