Modeling oncogenic signaling in colon tumors by multidirectional analyses of microarray data directed for maximization of analytical reliability.

Skrzypczak M, Goryca K, Rubel T, Paziewska A, Mikula M, Jarosz D, Pachlewski J, Oledzki J, Ostrowski J. · 2010

Read Modeling oncogenic signaling in colon tumors by multidirectional analyses of microarray data directed for maximization of analytical reliability. by Skrzypczak M, Goryca K, Rubel T, Paziewska A, Mikula M, Jarosz D, Pachlewski J, Oledzki J, Ostrowski J. (2010) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc2948500.txt.

Introduction

Skrzypczak and co-authors mapped signalling changes as bowel growths progress from normal lining to benign polyps to cancer, in a 2010 study combining several analysis routes. Single-method gene studies often disagree, so the team sought a more reliable consensus.

Samples from surgery and colonoscopy covered more than a hundred whole-tissue specimens plus micro-dissected subsets. Running two normalisation methods and several sorting rules in parallel, they kept only signals that survived every route. The consensus flagged sets of known pathway groups differing along the progression and between tissue layers, giving cleaner targets for follow-up work.

Catalog record

CREATORSkrzypczak M, Goryca K, Rubel T, Paziewska A, Mikula M, Jarosz D, Pachlewski J, Oledzki J, Ostrowski J.
YEAR2010
COLLECTIONPAPERS
FORMATTXT
FILE SIZE57 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC2948500

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Modeling Oncogenic Signaling in Colon Tumors by Multidirectional Analyses of Microarray Data Directed for Maximization of Analytical Reliability By Magdalena Skrzypczak, Krzysztof Goryca, Tymon Rubel, Agnieszka Paziewska, Michal Mikula, Dorota Jarosz, Jacek Pachlewski, Janusz Oledzki, Jerzy Ostrowsk, Raya Khanin Published in: PLoS ONE DOI: 10.1371/journal.pone.0013091 ABSTRACT Abstract Background Clinical progression of colorectal cancers (CRC) may occur in parallel with distinctive signaling alterations. We designed multidirectional analyses integrating microarray-based data with biostatistics and bioinformatics to elucidate the signaling and metabolic alterations underlying CRC development in the adenoma-carcinoma sequence. Methodology/Principal Findings Studies were performed on normal mucosa, adenoma, and carcinoma samples obtained during surgery or colonoscopy. Collections of cryostat sections prepared from the tissue samples were evaluated by a pathologist to control the relative cell type content. The measurements were done using Affymetrix GeneChip HG-U133plus2, and probe set data was generated using two normalization algorithms: MAS5.0 and GCRMA with least-variant set (LVS). The data was evaluated using pair-wise comparisons and data decomposition into singular value decomposition (SVD) modes. The method selected for the functional analysis used the Kolmogorov-Smirnov test. Expressional profiles obtained in 105 samples of whole tissue sections were used to establish oncogenic signaling alterations in progression of CRC, while those representing 40 microdissected specimens were used to select differences in KEGG pathways between epithelium and mucosa. Based on a consensus of the results obtained by two normalization algorithms, and two probe set sorting criteria, we identified 14 and 17 KEGG signaling and metabolic pathways that are significantly altered between normal and tumor samples and between benign and malignant tumors, respectively. Several of them were also selected from the raw microarray data of 2 recently published studies (GSE4183 and GSE8671). Conclusion/Significance Although the proposed strategy is computationally complex and labor–intensive, it may reduce the number of false results. INTRODUCTION Colorectal cancer (CRC) arises as a multi-step process of successive cellular clone selection. As a result of the growth advantage of dysplastic cells over their normal neighbors, the morphological counterpart of molecular alterations leads to progressive cytological and architectural derangement recognizable as the adenoma-carcinoma sequence [1], [2]. Recently, no more than a dozen or so somatic “driver” mutations were established as being responsible for CRC development [3], [4]. However, tumors exhibiting homogenous phenotypes share few mutated “cancer genes”; therefore, cancer complexity at the gene level is likely reduced to a limited number of alterations within signaling and metabolic pathways [5]. An individual cancer phenotype is the result of cell-specific, developmental stage-specific, and metabolism-related changes in gene expression selectively occurring at a time and modified by epigenetic interactions [6]. With the introduction of high-density DNA microarrays, an expectation of insight into the overall molecular components of carcinogenesis has developed. Unfortunately, a comparative analysis of microarray-based studies on CRC development found rather weak overlap of the gene expression profiles ([7] and the results section). These discrepancies in the identified expression profiles may be due to technical reasons, including the use of various microarray platforms, different tissue collection methods, and numerous analytical algorithms [7]. Although the rate of false assumptions might be minimized by using an “optimal” analytical protocol, the selection of such protocol is still challenging [6]. Consequently, microarray experiments allow for rough and mostly indirect assumptions. Therefore, one may ask whether multidimensional and sub-optimal microarray-based data can be applied to the study of complex biological systems, including carcinogenesis. To answer this question, microarray data originating from two experimental procedures were analyzed by multiple methods employed for identification of consensus differences in pathways underlying CRC development through the adenoma-carcinoma sequence. Samples of normal mucosa, adenomas, and carcinomas obtained during surgery or colonoscopy were processed to select differences between: (i) epithelium and mucosa in normal tissue away from and directly adjacent to carcinoma, in adenoma and carcinoma (using microdissected samples) and (ii) normal and neoplastic tissues and adenomas and carcinomas (using whole tissue sections). Finally, we addressed the potential and challenges of translating microarray-based gene expression profiles into the functional aspects of carcinogenesis. RESULTS A total of 170 GeneChips were hybridized in this study. Twenty-five of 130 arrays representing whole tissue section samples were rejected from the data analyses. Twenty-four of these arrays were rejected on the basis of poor GeneChip quality according to the parameters established by Affymetrix and due to their internal inconsistency with others as established by the principal component analysis (PCA) (not shown). One microarray was rejected because of mechanical damage. A parameter summary of 145 arrays of suitable quality, 40 and 105 represented microdissected and whole tissue section samples, respectively, is provided in Table S1. From 54,675 probe sets of the Affymetrix HGU133plus2 microarray, 31,962 and 25,410 probe sets for whole tissue sections and 29,242 and 24,002 probe sets for microdissected samples passed the filtering procedure according to MAS5.0 and GCRMA+ LVS algorithms, respectively. As shown in Figure S1, the probe set signal distribution and levels extracted with the two normalization algorithms significantly differed. Signaling pathways distinguishing between colonic epithelial cells and mucosa Probe set selection from microarray data representing tissue samples may be significantly affected by differences in the cell type content of the normal and dysplastic mucosa (Table 1). Bearing that in mind, we intended to define the consensus differences in KEGG pathways which are more conserved between colonic epithelial cells and mucosa than between normal and neoplastic tissues. Gene expression profiles corresponding to tissue morphology were established in pure colonic crypt epithelial cells (CEC) and mucosa (MUC) (representing the epithelial cell layer and cell content in lamina propria) which were captured from various parts of the tumor consisting of invasive adenocarcinoma adjacent to tubular adenoma with low grade dysplasia and from paired full-thickness normal colon using the laser capture microdissection. [Table: The relative cell type content within normal and dysplastic mucosa used for laser capture microdissection (per 1 mm2 of the area).] Data sets from these microdissected samples were analyzed in the following pairs: CEC vs. MUC dissected from distant full-thickness normal colon (NC), normal colon mucosa dissected from tumor (NT), adenoma (AD), and carcinoma (CA). Pair-wise comparisons were performed separately on microarray data sets normalized by MAS5.0 and GCRMA+LVS algorithms. Probe sets sorted according to significance of differentiation of CEC and MUC in any pair set were used in Kolmogorov-Smirnov (K-S) test to evaluate KEGG pathways alterations. The results of KEGG annotations for each comparison are summarized in Table S2. The pathways found in at least three of the four pair-wise comparisons (Table 2) were considered to be distinguishing between colonic epithelial cells and mucosa and were excluded from the further selection of oncogenic signaling pathways. [Table: KEGG pathways found significant (KS test) in pair-wise comparisons of pure colonic crypt epithelial cells (CEC) and mucosa (MUC) in at least three of four sample groups: normal colon (NC), normal tumor mucosa (NT), adenoma (AD), and carcinoma (CA).] Oncogenic signaling pathways To establish oncogenic signaling distinguishing between normal colon and neoplasms, data sets representing whole tissue sections of normal colon, adenomas, and adenocarcinomas and normalized by both MAS5.0 and GCRMA+LVS algorithms were decomposed into major statistically independent variability modes (supergenes) using singular value decomposition (SVD). We assumed that SVD may establish some co-variations in gene expression that could enable better definition of expression-based functional alterations [8]–[10]. A graphic

Read the complete text