Chemosensitivity prediction by transcriptional profiling

A mechanism of cyclin D1 action encoded in the patterns of gene expression in human cancer

Here we describe how patterns of gene expression in human tumors have been deconvoluted to reveal a mechanism of action for the cyclin D1 oncogene. Computational analysis of the expression patterns of thousands of genes across hundreds of tumor specimens suggested that a transcription factor, C/EBPbeta/Nf-Il6, participates in the consequences of cyclin D1 overexpression. Functional analyses confirmed the involvement of C/EBPbeta in the regulation of genes affected by cyclin D1 and established this protein as an indispensable effector of a potentially important facet of cyclin D1 biology. This work demonstrates that tumor gene expression databases can be used to study the function of a human oncogene in situ.

Mycosis fungoides shows concurrent deregulation of multiple genes involved in the TNF signaling pathway: an expression profile study

Mycosis fungoides (MF) is the most frequent type of cutaneous T-cell lymphoma, whose diagnosis and study is hampered by its morphologic similarity to inflammatory dermatoses (ID) and the low proportion of tumoral cells, which often account for only 5% to 10% of the total tissue cells. cDNA microarray studies using the CNIO OncoChip of 29 MF and 11 ID cases revealed a signature of 27 genes implicated in the tumorigenesis of MF, including tumor necrosis factor receptor (TNFR)-dependent apoptosis regulators, STAT4, CD40L, and other oncogenes and apoptosis inhibitors. Subsequently a 6-gene prediction model was constructed that is capable of distinguishing MF and ID cases with unprecedented accuracy. This model correctly predicted the class of 97% of cases in a blind test validation using 24 MF patients with low clinical stages. Unsupervised hierarchic clustering has revealed 2 major subclasses of MF, one of which tends to include more aggressive-type MF cases including tumoral MF forms. Furthermore, signatures associated with abnormal immunophenotype (11 genes) and tumor stage disease (5 genes) were identified.

MLANA/MART1 and SILV/PMEL17/GP100 are transcriptionally regulated by MITF in melanocytes and melanoma

The clinically important melanoma diagnostic antibodies HMB-45, melan-A, and MITF (D5) recognize gene products of the melanocyte-lineage genes SILV/PMEL17/GP100, MLANA/MART1, and MITF, respectively. MITF encodes a transcription factor that is essential for normal melanocyte development and appears to regulate expression of several pigmentation genes. In this report, the possibility was examined that MITF might additionally regulate expression of the SILV and MLANA genes. Both genes contain conserved MITF consensus DNA sequences that were bound by MITF in vitro and in vivo, based on electrophoretic mobility shift assay and chromatin-immunoprecipitation. In addition, MITF regulated their promoter/enhancer regions in reporter assays, and up- or down-regulation of MITF produced corresponding modulation of endogenous SILV and MLANA in melanoma cells. Expression patterns were compared with these factors in a series of melanoma cell lines whose mutational status of the proto-oncogene BRAF was also known. SILV and MLANA expression correlated with MITF, while no clear correlation was seen relative to BRAF mutation. Finally, mRNA expression array analysis of primary human melanomas demonstrated a tight correlation in their expression levels in clinical tumor specimens. Collectively, this study links three important melanoma antigens into a common transcriptional pathway regulated by MITF.

A rational approach to personalized anticancer therapy: chemoinformatic analysis reveals mechanistic gene-drug associations

PURPOSE

To predict the response of cells to chemotherapeutic agents based on gene expression profiles, we performed a chemoinformatic study of a set of standard anticancer agents assayed for activity against a panel of 60 human tumor-derived cell lines from the Developmental Therapeutics Program (DTP) at the National Cancer Institute (NCI).

METHODS

Mechanistically-relevant gene:drug activity associations were identified in the scientific literature. The correlations between expression levels of drug target genes and the activity of the drugs against the NCI's 60 cell line panel were calculated across and within each tumor tissue type, using published drug activity and gene expression data.

RESULTS

Compared to other mechanistically-relevant gene-drug associations, that of triciribine phosphate (TCN-P) and adenosine kinase (ADK) was exceptionally strong--overall and within tumor tissue types-across the 60 cell lines profiled for chemosensitivity (1) and gene expression (2).

CONCLUSION

The results suggest ADK expression may be useful for stratifying TCN-P-responsive vs. non-responsive tumors. Based on TCN-P's mechanism of action and the observed TCN-P:ADK association, we contend that catalytic drug activation provides a rational, mechanistic basis for personalizing cancer treatment based on tumor-specific differences in the expression of drug-activating enzymes.

Chemical genomics as an emerging paradigm for postgenomic drug discovery

Chemical genomics approaches are evolving to overcome key problems limiting the efficiency of drug discovery in the postgenomic era. Many of these stem from the low success rates in finding drugs for novel genomics targets whose biochemical properties and therapeutic relevance is poorly understood. The fundamental objective of chemical genomics is to find and optimize chemical compounds that can be used to directly test the therapeutic relevance of new targets revealed through genome sequencing. An integrated approach to chemical genomics encompasses a diverse set of tools including quantitative affinity-based screens, computer-directed combinatorial chemistry, and structure-based drug design. The approach is most effectively applied across targets classes whose members are structurally related, and where some members are known to have bona fide therapeutic relevance.

Drug sensitivity and resistance genes in cancer chemotherapy: a chemogenomics approach

Resistance to antineoplastic drugs represents a serious obstacle to successful cancer treatment. Genome-wide studies correlating drug response phenotypes with large DNA/tissue microarray and proteomic datasets have been performed to identify the genes and proteins involved in chemosensitivity or drug resistance. The goal is to identify a set of chemosensitivity and/or resistance genes for each drug that are predictive of treatment response. Therefore, validated pharmacogenomic biomarkers offer the potential for the selection of optimal treatment regimens for individual patients and for identifying novel therapeutic targets to overcome drug resistance.

Differential gene expression and genomic patient stratification following left ventricular assist device support

OBJECTIVES

We sought to determine whether mechanical unloading of the failing human heart with a left ventricular assist device (LVAD) results in significant changes in overall left ventricular gene expression.

BACKGROUND

Mechanical circulatory support by LVAD in end-stage human heart failure (HF) can result in beneficial reverse remodeling of myocardial structure and function. The molecular mechanisms behind this salutary process are not well understood.

METHODS

Left ventricular samples from six male patients were harvested during LVAD placement and subsequently at the time of explantation. Cardiac gene expression was determined using oligonucleotide microarrays.

RESULTS

Paired t test analysis revealed numerous genes that were regulated in a statistically significant fashion, including the downregulation of several previously studied genes. Further statistical analysis revealed that the overall gene expression profiles could significantly distinguish pre- and post-LVAD status. Interestingly, the data also identified two distinct groups among the pre-LVAD failing hearts, in which there was blind segregation of patients based on HF etiology. In addition to the substantial divergence in genomic profiles for these two HF groups, there were significant differences in their corresponding LVAD-mediated regulation of gene expression.

CONCLUSIONS

Support with an LVAD in HF induces significant changes in myocardial gene expression, as pre- and post-LVAD hearts demonstrate significantly distinct genomic footprints. Thus, reverse remodeling is associated with a specific pattern of gene expression. Moreover, we found that deoxyribonucleic acid microarray technology could distinguish, in a blind manner, patients with different HF etiologies. Expansion of this study and further development of these statistical methods may facilitate prognostic prediction of the individual patient response to LVAD support.

High-content profiling of drug-drug interactions: cellular targets involved in the modulation of microtubule drug action by the antifungal ketoconazole

Drug-drug interactions play an important role in the discovery and development of therapeutic agents. High-content profiling was developed to unravel the complexity of these interactions by providing multiparameter measurements of target activity at the cellular and subcellular levels. Two microtubule drugs, vinblastine and curacin A, were shown to modulate multiple cellular processes, including nuclear condensation, the activation of the extracellular signal-regulated kinase pathway as measured by RSK90 phosphorylation, and the regulation of the microtubule cytoskeleton as measured in detergent-extracted cells. The heterogeneity of the response, addressed through population analysis and multiparameter comparisons within single cells, was consistent with vinblastine and curacin A having similar effects on nuclear morphology and 90 kDa ribosomal s6 kinase (RSK90) phosphorylation despite having distinct effects on the microtubule cytoskeleton. Ketoconazole, originally developed as an antifungal agent, exhibited concentration-dependent inhibitory and potentiating effects on both drugs in HeLa and PC-3 cells at concentration ranges near the plasma levels of ketoconazole attained in human subjects. Thus, high-content profiling was used to dissect the cellular and molecular responses to interacting drugs and is therefore a potentially important tool in the selection, characterization, and optimization of lead therapeutic compounds.

DNA arrays in clinical oncology: promises and challenges

Cancer is a complex genetic disease characterized by the accumulation of multiple molecular alterations. Current diagnostic and prognostic classifications, based on clinical and pathologic factors, are insufficient to reflect the whole clinical heterogeneity of tumors. Most current anticancer agents do not differentiate between cancerous and normal cells, leading sometimes to disastrous adverse effects. Recent advances in human genome research and high-throughput molecular technologies make it possible finally to tackle the molecular complexity of malignant tumors. With DNA array technology, mRNA expression levels of thousands of genes can be measured simultaneously in a single assay. Oncology is benefiting on multiple fronts. Gene expression profiles are revealing new biologically and clinically relevant tumor subclasses previously indistinguishable and are identifying new diagnostic and prognostic biomarkers as well as new potential therapeutic targets. Here, we review the technology and present clinical applications for which promising results have been obtained. Finally, we discuss issues that must be resolved in the near future to allow DNA arrays to translate into benefits for cancer patients.

Iterative signature algorithm for the analysis of large-scale gene expression data

We present an approach for the analysis of genome-wide expression data. Our method is designed to overcome the limitations of traditional techniques, when applied to large-scale data. Rather than alloting each gene to a single cluster, we assign both genes and conditions to context-dependent and potentially overlapping transcription modules. We provide a rigorous definition of a transcription module as the object to be retrieved from the expression data. An efficient algorithm, which searches for the modules encoded in the data by iteratively refining sets of genes and conditions until they match this definition, is established. Each iteration involves a linear map, induced by the normalized expression matrix, followed by the application of a threshold function. We argue that our method is in fact a generalization of singular value decomposition, which corresponds to the special case where no threshold is applied. We show analytically that for noisy expression data our approach leads to better classification due to the implementation of the threshold. This result is confirmed by numerical analyses based on in silico expression data. We discuss briefly results obtained by applying our algorithm to expression data from the yeast Saccharomyces cerevisiae.

Pharmacogenomic analysis: correlating molecular substructure classes with microarray gene expression data

Genomic studies are producing large databases of molecular information on cancers and other cell and tissue types. Hence, we have the opportunity to link these accumulating data to the drug discovery processes. Our previous efforts at 'information-intensive' molecular pharmacology have focused on the relationship between patterns of gene expression and patterns of drug activity. In the present study, we take the process a step further-relating gene expression patterns, not just to the drugs as entities, but to approximately 27,000 substructures and other chemical features within the drugs. This coupling of genomic information with structure-based data mining can be used to identify classes of compounds for which detailed experimental structure-activity studies may be fruitful. Using a systematic substructure analysis coupled with statistical correlations of compound activity with differential gene expression, we have identified two subclasses of quinones whose patterns of activity in the National Cancer Institute's 60-cell line screening panel (NCI-60) correlate strongly with the expression patterns of particular genes: (i) The growth inhibitory patterns of an electron-withdrawing subclass of benzodithiophenedione-containing compounds over the NCI-60 are highly correlated with the expression patterns of Rab7 and other melanoma-specific genes; (ii) the inhibitory patterns of indolonaphthoquinone-containing compounds are highly correlated with the expression patterns of the hematopoietic lineage-specific gene HS1 and other leukemia genes. As illustrated by these proof-of-principle examples, we introduce here a set of conceptual tools and fluent computational methods for projecting directly from gene expression patterns to drug substructures and vice versa. The analysis is presented in terms of the NCI-60 cell lines and microarray-based gene expression patterns, but the concept and methods are broadly applicable to other large-scale pharmacogenomic database sets as well. The approach (SAT for Structure-Activity-Target) provides a systematic way to mine databases for the design of further structure-activity studies, particularly to aid in target and lead identification.

Transcriptomic classification of antitumor agents: application to the analysis of the antitumoral effect of SR31747A

SR31747A is a sigma ligand that exhibits a potent antitumoral activity on various human tumor cell lines both in vitro and in vivo. To understand its mode of action, we used DNA microarray technology combined with a new bioinformatic approach to identify genes that are modulated by SR31747A in different human breast or prostate cancer cell lines. The SR31747A transcriptional signature was also compared with that of seven different representative anticancer drugs commonly used in the clinic. To this aim, we performed a two-dimensional hierarchical clustering analysis of drugs and genes which showed that 1) standard molecules with similar mechanism of action clustered together and 2) SR31747A does not belong to any previously characterized class of standard anticancer drugs. Moreover, we showed that 3) SR31747A mainly exerted its antiproliferative effect by inhibiting the expression of genes playing a key role in DNA replication and cell cycle progression. Finally, contrasting with other drugs, we obtained evidence that 4) SR31747A strongly inhibited the expression of three key enzymes of the nucleotide synthesis pathway (i.e., dihydrofolate reductase, thymidylate synthase, and thymidine kinase) with the latter shown both at the mRNA and protein levels. These results, obtained through a novel molecular approach to characterize and compare anticancer agents, showed that SR31747A exhibits an original mechanism of action, very likely through unexpected targets whose modulations may account for its antitumoral effect.

Gene expression profiling of endocrine tumors using DNA microarrays: progress and promise

Genomewide profiling of gene expression, made possible by the development of DNA microarray technology and more powerful by the sequencing of the human genome, has led to advances in tumor classification and biomarker discovery for the common types of human neoplasia. Application of this approach to the field of endocrine neoplasia is in its infancy, although some progress has been recently reported. In this review, the progress to date is summarized and the promise of DNA microarray analysis in conjunction with tissue array immunohistochemistry to significantly impact endocrine tumor diagnosis and prognosis is discussed.

DNA microarray and cancer

Since the discovery of the first oncogene 26 years ago, a large body of research has convincingly demonstrated that the initiation and progression of cancers involve the accumulation of genetic aberrations in the cell. Many techniques have been developed to identify these genetic abnormalities. The recent completion of human genome sequencing and advances in DNA microarray technology allow rapid genetic analysis to take place on a genome-wide scale and have revolutionized the way cancers are studied. This ground-breaking approach of studying cancer promises to provide a better understanding of the underlying mechanism for tumorigenesis, more accurate diagnosis, more comprehensive prognosis, and more effective therapeutic interventions.

MatchMiner: a tool for batch navigation among gene and gene product identifiers

MatchMiner is a freely available program package for batch navigation among gene and gene product identifier types commonly encountered in microarray studies and other forms of 'omic' research. The user inputs a list of gene identifiers and then uses the Merge function to find the overlap with a second list of identifiers of either the same or a different type or uses the LookUp function to find corresponding identifiers.

Better therapeutics through microarrays

DNA microarrays are an integral part of the process for therapeutic discovery, optimization and clinical validation. At an early stage, investigators use arrays to prioritize a few genes as potential therapeutic targets on the basis of various criteria. Subsequently, gene expression analysis assists in drug discovery and toxicology by eliminating poor compounds and optimizing the selection of promising leads. Integral to this process is the use of sophisticated statistics, mathematics and bioinformatics to define statistically valid observations and to deduce complex patterns of phenotypes and biological pathways. In short, microarrays are redefining the drug discovery process by providing greater knowledge at each step and by illuminating the complex workings of biological systems.

From patterns to pathways: gene expression data analysis comes of age

Many different biological questions are routinely studied using transcriptional profiling on microarrays. A wide range of approaches are available for gleaning insights from the data obtained from such experiments. The appropriate choice of data-analysis technique depends both on the data and on the goals of the experiment. This review summarizes some of the common themes in microarray data analysis, including detection of differential expression, clustering, and predicting sample characteristics. Several approaches to each problem, and their relative merits, are discussed and key areas for additional research highlighted.

Yellow pages to the transcriptome

Transcriptomics has become an important tool for the large-scale analysis of biological processes. This review aims to provide sufficient criteria to make an appropriate choice among the variety of 'closed' systems, represented by DNA microarrays, and 'open' systems like fragment display, tag sequencing and subtractive hybridization, depending on the biological system under investigation. The most important technologies currently available are presented, their strengths and weaknesses are discussed and companies active in the field are listed. The potential of transcriptomics in the pharmaceutical research and development process is highlighted by applications in oncology, research on neurological diseases, and predictive toxicology. Finally, a prognosis for future developments of the technologies is given.

Expression profiling with oligonucleotide arrays: technologies and applications for neurobiology

DNA microarrays have been used in applications ranging from the assignment of gene function to analytical uses in prognostics. However, the detection sensitivity, cross hybridization, and reproducibility of these arrays can affect experimental design and data interpretation. Moreover, several technologies are available for fabrication of oligonucleotide microarrays. We review these technologies and performance attributes and, with data sets generated from human brain RNA, present statistical tools and methods to analyze data quality and to mine and visualize the data. Our data show high reproducibility and should allow an investigator to discern biological and regional variability from differential expression. Although we have used brain RNA as a model system to illustrate some of these points, the oligonucleotide arrays and methods employed in this study can be used with cell lines, tissue sections, blood, and other fluids. To further demonstrate this point, we provide data generated from total RNA sample sizes of 200 ng.

Novel strategies for overcoming multidrug resistance in cancer

The problems of why metastatic cancers develop pleiotropic resistance to all available therapies, and how this might be countered, are the most pressing in cancer chemotherapy. It is likely that such resistance involves a combination of mechanisms including changes in drug transport/drug targets, reduction in the degree of drug-induced apoptosis/cell loss, and increased rate of tumour repopulation following therapy. Current research must consider not only which mechanisms contribute, eventually relating this to individual patients with cancer, but also what strategies might be utilised to counter each of the important resistance mechanisms. A considerable amount of work has been devoted to the development of inhibitors of membrane-associated transport proteins such as P-glycoprotein, which mediate drug efflux. This work is now being complemented by approaches that target cell death pathways such as those mediated by release of mitochondrial proteins and by activation of surface receptors such as Fas. Rapid progress has been made in developing small-molecular-weight drugs that influence the rate of apoptosis, for instance by binding to the bcl-2 family of proteins regulating mitochondrial permeability. Antisense approaches aimed at reducing bcl-2 expression, and thus increasing the rate of cell death, are also showing promise. Modification of repopulation kinetics provides a further approach but has not received as much attention as other aspects of tumour resistance. New therapeutic approaches will have to be complemented by improved diagnostic tests to evaluate the contributions of different resistance mechanisms in individual patients with cancer.

Comprehensive analysis of the gene expression profiles in human gastric cancer cell lines

Gastric adenocarcinoma is one of the major malignancies worldwide. Gastric cell lines have been widely used as the model to study the genetics, pharmacology and biochemistry of gastric cancers. Here we describe a comprehensive survey of the gene expression profiles of 12 gastric carcinoma cell lines, using cDNA microarray with 43 000 clones. For comparison, we also explored the gene expression patterns of 15 cell lines derived from lymphoid, endothelial, stromal and other epithelial cancers. Expression levels of specific genes were validated through comparison to protein expression by immunohistochemistry using cell block arrays. We found sets of genes whose expression corresponds to the molecular signature of each cell type. In the gastric cancer cell lines, apart from genes that are highly expressed corresponding to their common epithelial origin from the gastrointestinal tract, we found marked heterogeneity among the gene expression patterns of these cell lines. Some of the heterogeneity may reflect their underlying molecular characteristics or specific differentiation program. Two putative gastric carcinoma cell lines were found to be B-cell lymphoma, and another one had no epithelial specific gene expression and hence was of doubtful epithelial origin. These cell lines should no longer be used in gastric carcinoma research. In conclusion, our gene expression database can serve as a powerful resource for the study of gastric cancer using these cell lines.

Microarrays in pharmacogenomics--advances and future promise

With their ability to provide global views of genome sequence and gene activity, microarrays have emerged as key analytical tools in the field of pharmacogenomics. Vast amounts of data must be collected and analyzed to meet pharmacogenomics' ambitious goals, ranging from identifying markers that predict individuals' responses to therapy to discovering new drug targets. Microarrays will be instrumental to these efforts because they provide bountiful sources of gene expression and genotypic data. Attesting to their productivity, microarrays have been the central technology used in thousands of peer-reviewed publications and have also become important contributors to many databases including PharmGKB, the Cancer Microarray Database and the database of single nucleotide polymorphisms (dbSNP). Microarrays are also making more focused contributions, however, in helping pursue hypothesis-driven inquiries that extend or complement broad genomic surveys. In addition, their potential as clinical tools is being increasingly recognized. This review identifies some of the varied and changing needs of pharmacogenomics research and discusses the ways in which microarrays are tending to these demands. The technique's strongpoints and limitations are examined, as well as its future potential.

cDNA microarray analysis of global gene expression in sarcomas

Sequencing of the human genome and rapidly evolving microarray technology have combined to provide investigators with the ability to analyze individual tumors and groups of tumors for global patterns of gene expression. Few of these types of studies have been performed on rhabdomyosarcomas and osteogenic sarcomas, including cell lines and animal models. Groups of expressed genes that may characterize rhabdomyosarcomas and their subgroups and separate them from other types of tumors have been identified. More specifically, genes involved in myogenesis or the inhibition of myogenesis have been identified, as have genes that may play a role in metastatic activity in osteogenic sarcomas. Also, a study documenting the consistent and specific gene expression profile of gastrointestinal stromal tumors has been published. While the data regarding gene expression patterns in sarcomas is accruing, numerous investigators are working on developing and enhancing bioinformatic skills and tools such that the vast amount of data can be converted into knowledge regarding biology, therapeutic responsiveness or resistance, and prognosis.

Global analysis of large-scale chemical and biological experiments

Research in the life sciences is increasingly dominated by high-throughput data collection methods that benefit from a global approach to data analysis. Recent innovations that facilitate such comprehensive analyses are highlighted. Several developments enable the study of the relationships between newly derived experimental information, such as biological activity in chemical screens or gene expression studies, and prior information, such as physical descriptors for small molecules or functional annotation for genes. The way in which global analyses can be applied to both chemical screens and transcription profiling experiments using a set of common machine learning tools is discussed.

Analyzing array data using supervised methods

Pharmacogenomics is the application of genomic technologies to drug discovery and development, as well as for the elucidation of the mechanisms of drug action on cells and organisms. DNA microarrays measure genome-wide gene expression patterns and are an important tool for pharmacogenomic applications, such as the identification of molecular targets for drugs, toxicological studies and molecular diagnostics. Genome-wide investigations generate vast amounts of data and there is a need for computational methods to manage and analyze this information. Recently, several supervised methods, in which other information is utilized together with gene expression data, have been used to characterize genes and samples. The choice of analysis methods will influence the results and their interpretation, therefore it is important to be familiar with each method, its scope and limitations. Here, methods with special reference to applications for pharmacogenomics are reviewed.

Mining functional relationships in feature subspaces from gene expression profiles and drug activity profiles

In an effort to determine putative functional relationships between gene expression patterns and drug activity patterns of 60 human cancer cell lines, a novel method was developed to discover local associations within cell line subsets. The association of drug-gene pairs is an explorative way of discovering gene markers that predict clinical tumor sensitivity to therapy. Nine drug-gene networks were discovered, as well as dozens of gene-gene and drug-drug networks. Three drug-gene networks with well studied members were discussed and the literature shows that hypothetical functional relationships exist. Therefore, this method enables the gathering of new information beyond global associations.

DNA microarrays in clinical oncology

Aberrant gene expression is critical for tumor initiation and progression. However, we lack a comprehensive understanding of all genes that are aberrantly expressed in human cancer. Recently, DNA microarrays have been used to obtain global views of human cancer gene expression and to identify genetic markers that might be important for diagnosis and therapy. We review clinical applications of these novel tools, discuss some important recent studies, identify promising avenues of research in this emerging field of study, and discuss the likely impact that expression profiling will have on clinical oncology.

Multiclass cancer diagnosis using tumor gene expression signatures

The optimal treatment of patients with cancer depends on establishing accurate diagnoses by using a complex combination of clinical and histopathological data. In some instances, this task is difficult or impossible because of atypical clinical presentation or histopathology. To determine whether the diagnosis of multiple common adult malignancies could be achieved purely by molecular classification, we subjected 218 tumor samples, spanning 14 common tumor types, and 90 normal tissue samples to oligonucleotide microarray gene expression analysis. The expression levels of 16,063 genes and expressed sequence tags were used to evaluate the accuracy of a multiclass classifier based on a support vector machine algorithm. Overall classification accuracy was 78%, far exceeding the accuracy of random classification (9%). Poorly differentiated cancers resulted in low-confidence predictions and could not be accurately classified according to their tissue of origin, indicating that they are molecularly distinct entities with dramatically different gene expression patterns compared with their well differentiated counterparts. Taken together, these results demonstrate the feasibility of accurate, multiclass molecular cancer classification and suggest a strategy for future clinical implementation of molecular cancer diagnostics.