Search

Human breast tumors are heterogeneous and consist of phenotypically diverse cells. Breast cancer cells with a CD44+/CD24- phenotype have been suggested to have tumor-initiating properties with stem cell-like and invasive features, although it is unclear whether their presence within a tumor has clinical implications. There is also a large heterogeneity between tumors, illustrated by reproducible stratification into various subtypes based on gene expression profiles or histopathological features. We have explored the prevalence of cells with different CD44/CD24 phenotypes within breast cancer subtypes. Double-staining immunohistochemistry was used to quantify CD44 and CD24 expression in 240 human breast tumors for which information on other tumor markers and clinical characteristics was available. Gene expression data were also accessible for a cohort of the material. A considerable heterogeneity in CD44 and CD24 expression was seen both between and within tumors. A complete lack of both proteins was evident in 35% of the tumors, while 13% contained cells of more than one of the CD44+/CD24-, CD44-/CD24+ and CD44+/CD24+ phenotypes. CD44+/CD24- cells were detected in 31% of the tumors, ranging in proportion from only a few to close to 100% of tumor cells. The CD44+/CD24- phenotype was most common in the basal-like subgroup – characterized as negative for the estrogen and progesterone receptors as well as for HER2, and as positive for cytokeratin 5/14 and/or epidermal growth factor receptor, and particularly common in BRCA1 hereditary tumors, of which 94% contained CD44+/CD24- cells. The CD44+/CD24- phenotype was surprisingly scarce in HER2+ tumors, which had a predominantly CD24+ status. A CD44+/CD24- gene expression signature was generated, which included CD44 and α6-integrin (CD49f) among the top-ranked overexpressed genes. We demonstrate an association between basal-like and particularly BRCA1 hereditary breast cancer and the presence of CD44+/CD24- cells. Not all basal-like tumors and very few HER2+ tumors, however, contain CD44+/CD24- cells, emphasizing that a putative tumorigenic ability may not be confined to cells of this phenotype and that other breast cancer stem cell markers remain to be identified.

In recent years, the advent of experimental methods to probe gene expression profiles of cancer on a genome-wide scale has led to widespread use of supervised machine learning algorithms to characterize these profiles. The main applications of these analysis methods range from assigning functional classes of previously uncharacterized genes to classification and prediction of different cancer tissues. This article surveys the application of machine learning algorithms to classification and diagnosis of cancer based on expression profiles. To exemplify the important issues of the classification procedure, the emphasis of this article is on one such method, namely artificial neural networks. In addition, methods to extract genes that are important for the performance of a classifier, as well as the influence of sample selection on prediction results are discussed.

The majority of microarray studies evaluate gene ex- pression differences between various specimens or con- ditions. However, the causes of this variability often re- main unknown. Our aim is to identify underlying causes of these patterns, a process that would eventually enable a mechanistic understanding of the deregulation of gene expression in cancer. The procedure consists of three phases: pre-processing, data integration and statistical analysis. We have applied the strategy to identify genes that are overexpressed due to amplification in breast cancer. The data were obtained from 14 breast cancer cell lines, which were subjected to cDNA microarray based copy number and expression experiments. The re- sult of the analysis was a list that consisted of 92 genes. This set includes several genes that are known to be both overexpressed and amplified in breast cancer. The com- plete study was published in Journal of the Franklin In- stitute 2004, and in this paper we focus on the main issues of the study.

We derive a partition function for the Lund fragmentation model and compare it with that of a classical gas. For a fixed rapidity ``volume'' this partition function corresponds to a multiplicity distribution which is very close to a binomial distribution. We compare our results with the multiplicity distributions obtained from the JETSET Monte Carlo for several scenarios. Firstly, for the fragmentation vertices of the Lund string. Secondly, for the final state particles both with and without decays., Latex, 21+1 pages, 11 figures

Background DNA methylation at CpG dinucleotides is modified in tumorigenesis with potential impact on transcriptional activity. Methods We used the Illumina 450 K platform to evaluate DNA methylation patterns of 50 metastatic melanoma tumors, with matched gene expression data. Results We identified three different methylation groups and validated the groups in independent data from The Cancer Genome Atlas. One group displayed hypermethylation of a developmental promoter set, genome-wide demethylation, increased proliferation and activity of the SWI/SNF complex. A second group had a methylation pattern resembling stromal and leukocyte cells, over-expressed an immune signature and had improved survival rates in metastatic tumors (p

Classification of expression profiles to predict disease characteristics of for example cancer is a common application in high-throughput gene and protein expression research. Cross-validation is often used to optimize design of classifiers, with the aim to construct an optimal single classifier. In this work, we explore if classification performance can be improved by aggregating classifiers into ensembles that use committee votes for classification.We investigated if combining classifiers into ensembles improved classification performance compared to single classifiers. A couple of commonly used classifiers, nearest centroid classifier and support vector machine, were evaluated using four publicly available data sets. We found ensemble methods generally performed betterthan corresponding single classifiers.

The prognostic and treatment-predictive markers currently in use for breast cancer are commonly based on the protein levels of individual genes (e.g., steroid receptors) or aspects of the tumor phenotype, such as histological grade and percentage of cells in the DNA synthesis phase of the cell cycle. Microarrays have previously been used to classify binary classes in breast cancer such as estrogen receptor (ER)-α status. To test whether the properties and specific values of conventional prognostic markers are encoded within tumor gene expression profiles, we have analyzed 48 well-characterized primary tumors from lymph node-negative breast cancer patients using 6728-element cDNA microarrays. In the present study, we used artificial neural networks trained with tumor gene expression data to predict the ER protein values on a continuous scale. Furthermore, we determined a gene expression profile-directed threshold for ER protein level to redefine the cutoff between ER-positive and ER-negative classes that may be more biologically relevant. With a similar approach, we studied the prediction of other prognostic parameters such as percentage cells in the S phase of the cell cycle (SPF), histological grade, DNA ploidy status, and progesterone receptor status. Interestingly, there was a consistent reciprocal relationship in expression levels of the genes important for both ER and SPF prediction. This and similar studies may be used to increase our understanding of the biology underlying these markers as well as to improve the currently available prognostic markers for breast cancer.

Background Today, no objective criteria exist to differentiate between individual primary tumors and intra- or intermammary dissemination respectively, in patients diagnosed with two or more synchronous breast cancers. To elucidate whether these tumors most likely arise through clonal expansion, or whether they represent individual primary tumors is of tumor biological interest and may have clinical implications. In this respect, high resolution genomic profiling may provide a more reliable approach than conventional histopathological and tumor biological factors. Methods 32 K tiling microarray-based comparative genomic hybridization (aCGH) was used to explore the genomic similarities among synchronous unilateral and bilateral invasive breast cancer tumor pairs, and was compared with histopathological and tumor biological parameters. Results Based on global copy number profiles and unsupervised hierarchical clustering, five of ten (p = 1.9 × 10-5) unilateral tumor pairs displayed similar genomic profiles within the pair, while only one of eight bilateral tumor pairs (p = 0.29) displayed pair-wise genomic similarities. DNA index, histological type and presence of vessel invasion correlated with the genomic analyses. Conclusion Synchronous unilateral tumor pairs are often genomically similar, while synchronous bilateral tumors most often represent individual primary tumors. However, two independent unilateral primary tumors can develop synchronously and contralateral tumor spread can occur. The presence of an intraductal component is not informative when establishing the independence of two tumors, while vessel invasion, the presence of which was found in clustering tumor pairs but not in tumor pairs that did not cluster together, supports the clustering outcome. Our data suggest that genomically similar unilateral tumor pairs may represent a more aggressive disease that requires the addition of more severe treatment modalities, and underscores the importance of evaluating the clonality of multiple tumors for optimal patient management. In summary, our findings demonstrate the importance of evaluating the properties of both tumors in order to determine the most optimal patient management.

Gastrointestinal stromal tumors (GISTs), the most common mesenchymal tumors of the digestive tract, are believed to arise from the interstitial cells of Cajal. GISTs are characterized by mutations in the proto-oncogene KIT that lead to constitutive activation of its tyrosine kinase activity. The tyrosine kinase inhibitor STI571, active against the BCR-ABL fusion protein in chronic myeloid leukemia, was recently shown to be highly effective in GISTs. We used 13,826-element cDNA microarrays to define the expression patterns of 13 KIT mutation-positive GISTs and compared them with the expression profiles of a group of spindle cell tumors from locations outside the gastrointestinal tract. Our results showed a remarkably distinct and uniform expression profile for all of the GISTs. In particular, hierarchical clustering of a subset of 113 cDNAs placed all of the GIST samples into one branch, with a Pearson correlation0.91. This homogeneity suggests that the molecular pathogenesis of a GIST results from expansion of a clone that has acquired an activating mutation in KIT without the extreme genetic instability found in the common epithelial cancers. The results provide insight into the histogenesis of GIST and the clinical behavior of this therapeutically responsive tumor.

Principal component analysis is often incorporated into genome-wide expression studies, but what is it and how can it be used to explore high-dimensional data?

We discuss what happens at the end of the QCD cascades. We show that, with just a few reasonable assumptions, the emission of soft gluons is constrained to produce an ordered field in the form of a helix. We describe how to modify the Lund fragmentation scheme in order to fragment such a field. Our modified fragmentation scheme yields results which are consistent with current experimental measurements, but predicts at least one signature which should be observable., 21+1 pages

Breast tumors from BRCA1 germ line mutation carriers typically exhibit features of the basal-like molecular subtype. However, the specific genes recurrently mutated as a consequence of BRCA1 dysfunction have not been fully elucidated. In this study, we used gene expression profiling to molecularly subtype 577 breast tumors, including 73 breast tumors from BRCA1/2 mutation carriers. Focusing on the RB1 locus, we analyzed 33 BRCA1-mutated, 36 BRCA2-mutated, and 48 non-BRCA1/2–mutated breast tumors using a custom-designed high-density oligomicroarray covering the RB1 gene. We found a strong association between the basal-like subtype and BRCA1-mutated breast tumors and the luminal B subtype and BRCA2-mutated breast tumors. RB1 was identified as a major target for genomic disruption in tumors arising in BRCA1 mutation carriers and in sporadic tumors with BRCA1 promoter methylation but rarely in other breast cancers. Homozygous deletions, intragenic breaks, or microdeletions were found in 33% of BRCA1-mutant tumors, 36% of BRCA1 promoter–methylated basal-like tumors, 13% of non-BRCA1–deficient basal-like tumors, and 3% of BRCA2-mutated tumors. In conclusion, RB1 was frequently inactivated by gross gene disruption in BRCA1 hereditary breast cancer and BRCA1-methylated sporadic basal-like breast cancer but rarely in BRCA2 hereditary breast cancer and non-BRCA1–deficient sporadic breast cancers. Together, our findings show the existence of genetic heterogeneity within the basal-like breast cancer subtype that is based upon BRCA1 status. Cancer Res; 72(16); 4028–36. ©2012 AACR.

The development of new drugs against cancer requires established cell lines. They are needed for in vitro studies to identify candidate drugs and in xenograft models to measure drug efficacy in vivo. Specific criteria need to be fulfilled by cell lines used in the evaluation of potential novel therapeutic agents. It is imperative that they display the features of the particular cancer under investigation. Given the documented heterogeneity of cancers, relevant subtypes need to be represented. In this study, we have examined these aspects for pediatric acute lymphoblastic leukemia. A panel of 13 leukemia cell lines recently established in our laboratory was analyzed. We used cDNA microarrays to define the gene expression profiles and compared the data with immunophenotyping and cytogenetic analyses. The expression profiles obtained showed excellent concordance with corresponding protein levels. Importantly, the panel of lines displayed the critical genetic features identified in clinically important acute lymphoblastic leukemia subtypes in childhood leukemia patients.