Microarrays and Crohn's disease: Collecting reliable information

C. Csillag,O. Nielsen,R. Borup,F. Nielsen

DOI: https://doi.org/10.1080/00365520510012226

2005-04-01

Scandinavian Journal of Gastroenterology

Abstract:A single microarray experiment allows thousands of microspots of DNA, protein, tissue or various other organic compounds (known as targets) to be quantified by the incidence of interactions with known binding ligands (known as probes), which are arrayed on a solid surface. Probes can be nucleic sequences, proteins, tissues, cells or other molecules. This setup allows unprecedented high-output parallel processing of biological interactions. It is not uncommon to experience the information generated by a well-designed microarray-based study promoting a hypothesis-generating paradigm [1]. This paradigm derives from the enormous amount of data created by the parallel hybridization of thousands of sequences based on a single microarray chip and from powerful computational algorithms aimed at identifying patterns of differential gene expression profiling or protein binding. Another field expected to undergo profound changes is that of clinical investigations aimed at biomarker discovery. While current models require hundreds of patients for statistical validation of a single biomarker, microarray-based setups will allow analysis of thousands of variables based on molecules extracted from relatively few patients. Although such a strategy raises concerns of a statistical nature [2], an advantage is that this model of small trials with a high density of data will allow precise patient stratification in accordance with the expression profile [3]. Efforts aimed at discovery of previously unrecognized disease subgroups based on differential gene expression profiles have already produced results for multiple myeloma [4] and lymphoma [5], and in various studies it has been possible to assign a particular gene expression profile to known clinically relevant disease classes, such as in breast cancer [6], chronic lymphocytic leukaemia [7], acute myeloid leukaemia [8,9], lymphomas [10], bladder cancer, [11], and Mycobacterium leprae infection [12]. DNA, microarray techniques might additionally be used to identify genetic determinants that affect response to drug therapy [13,14]. Crucial for the intepretation of these data is their reliable integration. Expression profiles from several experiments must be compatible, adequate annotation must be present, and procedures must be well described. Several ongoing efforts aim at standardization and storage of microarray data. However, there is a growing body of evidence suggesting that microarray data might lack reliability. Experimental variations and different methods for data integration yield results that vary more than might be predicted. This article reviews the first published microarray results regarding Crohn’s disease (CD) and intestinal physiology obtained through microarray approaches. Emphasis is given in results that point to an unexpected discrepancy in data. Attempts to minimize experimental and interpretational variation are also presented, as well as a discussion that tries to justify caution in assessing microarray-generated data.