Amanda Blackford, Marianna Zahurak, Adam Mamelak, Jeanne Kowalski, Daniel Sauder

Amanda Blackford

Nested, Non-Parametric, Correlative Analysis of Microarrays for heterogeneous Phenotype Characterization

International Conference on Analysis of Genomic Data

May 10-11, 2004

We present a nonparametric approach for selecting candidate genes to characterize a heterogeneous phenotype, such as Basal Cell Carcinoma (BCC) of the skin, that are nested among genes selected on the basis of their individual, similar effects upon an array-wide closeness measure. In this setting, a goal is to obtain a reliable characterization of phenotypes based on very high-dimensional data from a few samples. The proposed measure defines closeness based on gene signal profiles (functionals), rather than on isolated (numerical) differences in individual genes, between samples. Based on this measure, we successively examine the significance of the following: a set of similarly behaved genes relative to all arrayed genes, a set of candidate, relative to similarly behaved genes, individual candidate genes relative to non-candidates, and the direction, as over- or under-expressed, of candidate genes. In each setting, sample pairs are the units of analysis, with U-statistics the theoretical framework. We illustrate the method on a two-channel microarray experiment in which individual arrays were performed on BCC and patient-matched, normal skin specimens. To date, numerous mechanisms have been implicated in the development of BCC, including lifetime ultraviolet (UV) exposure, aberrations in tumor suppressor genes, ultraviolet-induced DNA mutations, DNA mismatch repair defects, immune dysregulation and free radical induction. The combination of environmental and host mechanisms implicated in BCC development, along with the histological diversity exhibited by such tumors, results in neoplasms with a somewhat heterogeneous genetic composition. To elucidate genes dysregulated in BCC, while addressing the inherent heterogeneity displayed by them, we applied our method to a two-channel microarray experiment to select candidate genes as commonly (among all tumors), differentially expressed between BCC and patient-matched normal skin specimens. Based on this analysis, several genes in the oxidative phosphorylation pathway were selected, lending support to the hypothesis that irregularities in normal mitochondrial function are involved in neoplastic disease.

Adam J. Mamelak, Jonathan Alder, Jeanne Kowalski, Donna Bilu, Elliot Weiss, Hideaki Watanabe, Irwin Freed, Binghe Wang, Eric Vonderheid, Daniel N. Sauder

Adam Mamelak

Comparative microarray analysis of the patch and plaque stages of Mycosis Fungoides

International Investigative Dermatology

April 30 – May 4, 2003

Mycosis Fungoides (MF) is the most common of a heterogeneous group of malignant T-cell lymphomas, formally known as Cutaneous T Cell Lymphoma. MF is traditionally thought of as a clonal CD4+ memory T cell disorder primarily manifesting in the skin. MF has a variable presentation with overlap between clinical stages and in its initial stages, is difficult to distinguish from other inflammatory diseases. In the present study, we sought to delineate genomic differences between initial presentation and later disease. We hypothesized that developing unique gene expression profiles for each clinical stage of this malignancy could directly aid in diagnostics, while providing insight into overall disease classification, prognosis and pathogenic mechanisms underlying this disorder. We utilized cDNA microarray technology to construct gene expression profiles from skin biopsies taken from patch and plaque stage MF patients. A 1717 cDNA microarray containing characterized human genes relevant to immunity and cancer was used for differential gene expression analysis of patch versus plaque MF lesions. The data obtained was subjected a non-parametric, nested analysis, which selected 20 genes as potentially differentially expressed. Among the candidate genes identified was an anti-oxidant protein, a member of the RAS oncogene family, and a xenobiotic-metabolizing gene involved in the metabolism of carcinogens. Susceptibility to cancer is frequently a pathological consequence of extensive oxyradical damage that leads to a cycle of cell death and regeneration and causes mutations in cancer-related genes. This data could suggest a role for oxidative stress and oxyradical overload in the development and pathogenesis of MF, as observed in other types of tumors.

Min Wang, Weiyan Cai, Jeanne Kowalski, William Markesbery

Min Wang

Altered Gene Expression of Hippocampal Tissue Samples in Patients with Alzheimer’s Disease

Society for Neurosciences 32^nd Annual Meeting

November 2-7, 2002

BACKGROUND

Alzheimer's disease (AD) is the most common form of dementia. The pathogenesis of this disease is poorly understood but the development of amyloid plaques and neurofibrillary tangles in the brain. Identifying the genes that altered in expression in Alzheimer's disease brain will facilitate better understanding the mechanism and improve the early detection, treatment and prevention of the disease.

METHOD

Short post-mortem (within 4 hours) hippocampal tissue samples of six AD patients (Braak stage VI/VI) and six normal cognitive control subjects (Braak stage I-II/VI) were provided by the Sanders Brown Center on Aging (Lexington, KY). Total RNA was extracted from liquid nitrogen frozen tissue samples. The expression of 1,176 genes was analyzed using Atlas Human 1.2 Arrays (Clontech Inc., Palo Alto, CA, USA). The array membranes were exposed on phosphor imaging screens, and scanned using PhosphorImager (Molecular Dydamics, Sunnyvale, CA, USA). The signal intensities of expressed genes were then obtained and analyzed using AtlasImage software from Clontech. The array intensity profiles and differential intensity between tissues were analyzed based upon a novel, correlation-type approach. RESULT Among the six AD hippocampus, 340 genes (30%) were identified as having similar intensity profiles, while among the six normal control tissue samples, 410 genes (35%) showed similar intensity patterns. Among the genes regarded as similar in intensity for within AD and normal tissue, 207 genes were common in both. Within these 207 genes, 18 were identified as potentially differentially expressed between AD and normal control tissues. Among these, the majority of genes were classified as transcription factors and signaling protein according to their functions. ETR103 (transcription factor), poly (ADP-ribose) polymerase (apoptosis associated protein), and neurogranin/RC3 (calcium-binding and signaling protein) were reported in the previous studies to be involved in the development of AD.

CONCLUSION

This study revealed 207 genes that shared similar intensity patterns, with 18 genes identified as potentially different in intensity between AD hippocampus and normal tissue samples. Further studies based upon these results are warranted for their validation, along with comparison of results from other approaches to analysis.

CAM Home

Applications

Papers

Oncology Biostatistics Home