Trond Hellem Bø

Microarray experiments generate data sets with information on the expression levels of thousands of genes in a set of biological samples. Unfortunately, such experiments often produce multiple missing expression values, normally due to various experimental problems. As many algorithms for gene expression analysis require a complete data matrix as input, the missing values have to be estimated in order to analyze the available data. Alternatively, genes and arrays can be removed until no missing values remain. However, for genes or arrays with only a small number of missing values, it is desirable to impute those values. For the subsequent analysis to be as informative as possible, it is essential that the estimates for the missing gene expression values are accurate. A small amount of badly estimated missing values in the data might be enough for clustering methods, such as hierachical clustering or K-means clustering, to produce misleading results. Thus, accurate methods for missing value estimation are needed. We present novel methods for estimation of missing values in microarray data sets that are based on the least squares principle, and that utilize correlations between both genes and arrays. For this set of methods, we use the common reference name LSimpute. We compare the estimation accuracy of our methods with the widely used KNNimpute on three complete data matrices from public data sets by randomly knocking out data (labeling as missing). From these tests, we conclude that our LSimpute methods produce estimates that consistently are more accurate than those obtained using KNNimpute. Additionally, we examine a more classic approach to missing value estimation based on expectation maximization (EM). We refer to our EM implementations as EMimpute, and the estimate errors using the EMimpute methods are compared with those our novel methods produce. The results indicate that on average, the estimates from our best performing LSimpute method are at least as accurate as those from the best EMimpute algorithm.

Although 75% of endometrial cancers are treated at an early stage, 15% to 20% of these recur. We performed an integrated analysis of genome-wide expression and copy-number data for primary endometrial carcinomas with extensive clinical and histopathological data to detect features predictive of recurrent disease. Unsupervised analysis of the expression data distinguished 2 major clusters with strikingly different phenotypes, including significant differences in disease-free survival. To identify possible mechanisms for these differences, we performed a global genomic survey of amplifications, deletions, and loss of heterozygosity, which identified 11 significantly amplified and 13 significantly deleted regions. Amplifications of 3q26.32 harboring the oncogene PIK3CA were associated with poor prognosis and segregated with the aggressive transcriptional cluster. Moreover, samples with PIK3CA amplification carried signatures associated with in vitro activation of PI3 kinase (PI3K), a signature that was shared by aggressive tumors without PIK3CA amplification. Tumors with loss of PTEN expression or PIK3CA overexpression that did not have PIK3CA amplification also shared the PI3K activation signature, high protein expression of the PI3K pathway member STMN1, and an aggressive phenotype in test and validation datasets. However, mutations of PTEN or PIK3CA were not associated with the same expression profile or aggressive phenotype. STMN1 expression had independent prognostic value. The results affirm the utility of systematic characterization of the cancer genome in clinically annotated specimens and suggest the particular importance of the PI3K pathway in patients who have aggressive endometrial cancer.

BACKGROUND: Methods for extracting useful information from the datasets produced by microarray experiments are at present of much interest. Here we present new methods for finding gene sets that are well suited for distinguishing experiment classes, such as healthy versus diseased tissues. Our methods are based on evaluating genes in pairs and evaluating how well a pair in combination distinguishes two experiment classes. We tested the ability of our pair-based methods to select gene sets that generalize the differences between experiment classes and compared the performance relative to two standard methods. To assess the ability to generalize class differences, we studied how well the gene sets we select are suited for learning a classifier. RESULTS: We show that the gene sets selected by our methods outperform the standard methods, in some cases by a large margin, in terms of cross-validation prediction accuracy of the learned classifier. We show that on two public datasets, accurate diagnoses can be made using only 15-30 genes. Our results have implications for how to select marker genes and how many gene measurements are needed for diagnostic purposes. CONCLUSION: When looking for differential expression between experiment classes, it may not be sufficient to look at each gene in a separate universe. Evaluating combinations of genes reveals interesting information that will not be discovered otherwise. Our results show that class prediction can be improved by taking advantage of this extra information.

We have characterized several stages of normal human B cell development in adult bone marrow by gene expression profiling of hemopoietic stem cells, early B (E-B), pro-B, pre-B, and immature B cells, using RNA amplification and Lymphochip cDNA microarrays (n = 6). Hierarchical clustering of 758 differentially expressed genes clearly separated the five populations. We used gene sets to investigate the functional assignment of the differentially expressed genes. Genes involved in VDJ recombination as well as B lineage-associated transcription factors (TCF3 (E2A), EBF, BCL11A, and PAX5) were turned on in E-B cells, before acquisition of CD19. Several transcription factors with unknown roles in B lymphoid cells demonstrated interesting expression patterns, including ZCCHC7 and ZHX2. Compared with hemopoietic stem cells and pro-B cells, E-B cells had increased expression of 18 genes, and these included IGJ, IL1RAP, BCL2, and CD62L. In addition, E-B cells expressed T/NK lineage and myeloid-associated genes including CD2, NOTCH1, CD99, PECAM1, TNFSF13B, and MPO. Expression of key genes was confirmed at the protein level by FACS analysis. Several of these Ags were heterogeneously expressed, providing a basis for further subdivision of E-B cells. Altogether, these results provide new information regarding expression of genes in early stages of human B cell development.

// Geir Bredholt 1, * , Monica Mannelqvist 1, * , Ingunn M. Stefansson 1, 2 , Even Birkeland 1 , Trond Hellem Bø 1, 3 , Anne M. Øyan 4, 5 , Jone Trovik 5, 6 , Karl-Henning Kalland 4, 5 , Inge Jonassen 3 , Helga B. Salvesen 5, 6 , Elisabeth Wik 1, 2 , Lars A. Akslen 1, 2 1 Centre for Cancer Biomarkers CCBIO, Department of Clinical Medicine, Section for Pathology, University of Bergen, Bergen, Norway 2 Department of Pathology, Haukeland University Hospital, Bergen, Norway 3 CCBIO, Department of Informatics, University of Bergen, Bergen, Norway 4 Department of Microbiology, Haukeland University Hospital, Bergen, Norway 5 Center for Cancer Biomarkers CCBIO, Department of Clinical Science, University of Bergen, Norway 6 Department of Gynecology and Obstetrics, Haukeland University Hospital, Bergen, Norway * These authors have contributed equally to this work Correspondence to: Lars A. Akslen, e-mail: lars.akslen@uib.no Keywords: necrosis, hypoxia, angiogenesis, inflammation, gene signatures Received: June 11, 2015      Accepted: August 26, 2015      Published: October 14, 2015 ABSTRACT Aims: Tumor necrosis is associated with aggressive features of endometrial cancer and poor prognosis. Here, we investigated gene expression patterns and potential treatment targets related to presence of tumor necrosis in primary endometrial cancer lesions. Methods and Results: By DNA microarray analysis, expression of genes related to tumor necrosis reflected multiple tumor-microenvironment interactions like tissue hypoxia, angiogenesis and inflammation pathways. A tumor necrosis signature of 38 genes and a related patient cluster (Cluster I, 67% of the cases) were associated with features of aggressive tumors such as type II cancers, estrogen receptor negative tumors and vascular invasion. Further, the tumor necrosis signature was increased in tumor cells grown in hypoxic conditions in vitro . Multiple genes with increased expression are known to be activated by HIF1A and NF-kB. Conclusions: Our findings indicate that the presence of tumor necrosis within primary tumors is associated with hypoxia, angiogenesis and inflammation responses. HIF1A, NF-kB and PI3K/mTOR might be potential treatment targets in aggressive endometrial cancers with presence of tumor necrosis.