Trond Hellem Bø

New feature subset selection procedures for classification of expression profiles

BackgroundMethods 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.ResultsWe 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.ConclusionWhen 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.

Integrated genomic profiling of endometrial carcinoma associates aggressive tumors with indicators of PI3 kinase activation

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.

Characterization of early stages of human B cell development by gene expression profiling.

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.

Tumor classification and marker gene prediction by feature selection and fuzzy c-means clustering using microarray data

BackgroundUsing DNA microarrays, we have developed two novel models for tumor classification and target gene prediction. First, gene expression profiles are summarized by optimally selected Self-Organizing Maps (SOMs), followed by tumor sample classification by Fuzzy C-means clustering. Then, the prediction of marker genes is accomplished by either manual feature selection (visualizing the weighted/mean SOM component plane) or automatic feature selection (by pair-wise Fishers linear discriminant).ResultsThe proposed models were tested on four published datasets: (1) Leukemia (2) Colon cancer (3) Brain tumors and (4) NCI cancer cell lines. The models gave class prediction with markedly reduced error rates compared to other class prediction approaches, and the importance of feature selection on microarray data analysis was also emphasized.ConclusionsOur models identify marker genes with predictive potential, often better than other available methods in the literature. The models are potentially useful for medical diagnostics and may reveal some insights into cancer classification. Additionally, we illustrated two limitations in tumor classification from microarray data related to the biology underlying the data, in terms of (1) the class size of data, and (2) the internal structure of classes. These limitations are not specific for the classification models used.

Increased Expression of SIM2-s Protein Is a Novel Marker of Aggressive Prostate Cancer

Purpose: The human SIM2 gene is located within the Downs syndrome critical region of chromosome 21 and encodes transcription factors involved in brain development and neuronal differentiation. SIM2 has been assigned a possible role in the pathogenesis of solid tumors, and the SIM2-short isoform (SIM2-s) was recently proposed as a molecular target for cancer therapy. We previously reported SIM2 among the highly up-regulated genes in 29 prostate cancers, and the purpose of our present study was to examine the expression status of SIM2 at the transcriptional and protein level as related to outcome in prostate cancer. Experimental Design: By quantitative PCR, mRNA in situ hybridization, and immunohistochemistry, we evaluated the expression and significance of SIM2 isoforms in 39 patients with clinically localized prostate cancer and validated the expression of SIM2-s protein in an independent cohort of 103 radical prostatectomies from patients with long and complete follow-up. Results: The SIM2 isoforms (SIM2-s and SIM2-l) were significantly coexpressed and increased in prostate cancer. Tumor cell expression of SIM2-s protein was associated with adverse clinicopathologic factors like increased preoperative serum prostate-specific antigen, high histologic grade, invasive tumor growth with extra-prostatic extension, and increased tumor cell proliferation by Ki-67 expression. SIM2-s protein expression was significantly associated with reduced cancer-specific survival in multivariate analyses. Conclusions: These novel findings indicate for the first time that SIM2 expression might be important for clinical progress of human cancer and support the recent proposal of SIM2-s as a candidate for targeted therapy in prostate cancer.

Gene expression patterns related to vascular invasion and aggressive features in endometrial cancer.

The presence of tumor cells entering vascular channels is a prognostic marker for many cancers, including endometrial carcinoma. Vascular invasion is considered to be an early step in the metastatic process and important for the progress of malignant tumors. Here, we investigated the gene expression patterns related to vascular involvement in 57 primary endometrial cancers, using DNA microarray and quantitative PCR techniques. A vascular invasion signature of 18 genes was significantly associated with patient survival and clinicopathological phenotype. Vascular involvement was also related to gene sets for epithelial-mesenchymal transition, wound response, endothelial cells, and vascular endothelial growth factor (VEGF) activity. With immunohistochemical validation, both collagen 8 and matrix metalloproteinase 3 (MMP3) were associated with vascular invasion, whereas ANGPTL4 and IL-8 were associated with patient survival. Our findings indicate that vascular involvement within primary tumors is associated with gene expression profiles related to angiogenesis and epithelial-mesenchymal transition. These data could contribute to an improved understanding of potential targets for metastatic spread and may provide clinically important information for better management of endometrial cancer.

CD34 expression in native human acute myelogenous leukemia blasts: Differences in CD34 membrane molecule expression are associated with different gene expression profiles

The stem cell marker CD34 is expressed by leukemia blasts only for a subset of patients with acute myelogenous leukemia (AML). It is still controversial as to whether CD34 expression (defined as at least 10–20% positive cells) has any prognostic effect in patients with AML who receive intensive chemotherapy. The present study investigated whether gene expression profiling could be used to further subclassify CD34+ AML cell populations.

Genomic Diversity among Beijing and non-Beijing Mycobacterium tuberculosis Isolates from Myanmar

Background The Beijing family of Mycobacterium tuberculosis is dominant in countries in East Asia. Genomic polymorphisms are a source of diversity within the M. tuberculosis genome and may account for the variation of virulence among M. tuberculosis isolates. Till date there are no studies that have examined the genomic composition of M. tuberculosis isolates from the high TB-burden country, Myanmar. Methodology/Principle Findings Twenty-two M. tuberculosis isolates from Myanmar were screened on whole-genome arrays containing genes from M. tuberculosis H37Rv, M. tuberculosis CDC1551 and M. bovis AF22197. Screening identified 198 deletions or extra regions in the clinical isolates compared to H37Rv. Twenty-two regions differentiated between Beijing and non-Beijing isolates and were verified by PCR on an additional 40 isolates. Six regions (Rv0071-0074 [RD105], Rv1572-1576c [RD149], Rv1585c-1587c [RD149], MT1798-Rv1755c [RD152], Rv1761c [RD152] and Rv0279c) were deleted in Beijing isolates, of which 4 (Rv1572-1576c, Rv1585c-1587c, MT1798-Rv1755c and Rv1761c) were variably deleted among ST42 isolates, indicating a closer relationship between the Beijing and ST42 lineages. The TbD1 region, Mb1582-Mb1583 was deleted in Beijing and ST42 isolates. One M. bovis gene of unknown function, Mb3184c was present in all isolates, except 11 of 13 ST42 isolates. The CDC1551 gene, MT1360 coding for a putative adenylate cyclase, was present in all Beijing and ST42 isolates (except 1). The pks15/1 gene, coding for a putative virulence factor, was intact in all Beijing and non-Beijing isolates, except in ST42 and ST53 isolates. Conclusion This study describes previously unreported deletions/extra regions in Beijing and non-Beijing M. tuberculosis isolates. The modern and highly frequent ST42 lineage showed a closer relationship to the hypervirulent Beijing lineage than to the ancient non-Beijing lineages. The pks15/1 gene was disrupted only in modern non-Beijing isolates. This is the first report of an in-depth analysis on the genomic diversity of M. tuberculosis isolates from Myanmar.

Tumor necrosis is an important hallmark of aggressive endometrial cancer and associates with hypoxia, angiogenesis and inflammation responses.

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.

Genes of cell-cell interactions, chemotherapy detoxification and apoptosis are induced during chemotherapy of acute myeloid leukemia

BackgroundThe molecular changes in vivo in acute myeloid leukemia cells early after start of conventional genotoxic chemotherapy are incompletely understood, and it is not known if early molecular modulations reflect clinical response.MethodsThe gene expression was examined by whole genome 44 k oligo microarrays and 12 k cDNA microarrays in peripheral blood leukocytes collected from seven leukemia patients before treatment, 2–4 h and 18–24 h after start of chemotherapy and validated by real-time quantitative PCR. Statistically significantly upregulated genes were classified using gene ontology (GO) terms. Parallel samples were examined by flow cytometry for apoptosis by annexin V-binding and the expression of selected proteins were confirmed by immunoblotting.ResultsSignificant differential modulation of 151 genes were found at 4 h after start of induction therapy with cytarabine and anthracycline, including significant overexpression of 31 genes associated with p53 regulation. Within 4 h of chemotherapy the BCL2/BAX and BCL2/PUMA ratio were attenuated in proapoptotic direction. FLT3 mutations indicated that non-responders (5/7 patients, 8 versus 49 months survival) are characterized by a unique gene response profile before and at 4 h. At 18–24 h after chemotherapy, the gene expression of p53 target genes was attenuated, while genes involved in chemoresistance, cytarabine detoxification, chemokine networks and T cell receptor were prominent. No signs of apoptosis were observed in the collected cells, suggesting the treated patients as a physiological source of pre-apoptotic cells.ConclusionPre-apoptotic gene expression can be monitored within hours after start of chemotherapy in patients with acute myeloid leukemia, and may be useful in future determination of therapy responders. The low number of patients and the heterogeneity of acute myeloid leukemia limited the identification of gene expression predictive of therapy response. Therapy-induced gene expression reflects the complex biological processes involved in clinical cancer cell eradication and should be explored for future enhancement of therapy.