From Faulty Biochemical Assumptions to Poor Diagnostic Performance: Reflections on the Failure of DNA Microarrays to Become a Reliable Clinical Diagnostic Tool

Since their inception about two decades ago, DNA microarrays have been considered as a great hope in translational research and personalized medicine. Although DNA microarrays for gene expression profiling proved to be an indispensable tool in the laboratory settings, their applications as an instrument for clinical diagnostics have not yet produced tangible results. In this paper, we convey the idea that, apart from notoriously poor reproducibility and complexities of experimental validations, there exist other reasons hindering clinical application of DNA microarrays. These reasons are rooted in the very core of the DNA microarrays methodology, that is, in faulty biochemical assumptions underlying microarray measurements. A key premise the microarray measurements are based on is that mRNA abundances harvested from the eukaryotic cytoplasm are indicative of the activity levels of corresponding genes. There are at least two reasons why this premise is questionable. First, each transcription is supported by a number of transcription factors expressed by many genes. Due to this reason, relations between the transcription rates of genes and the mRNA abundances are the 'many-to-one', not the 'one-to-one'; therefore, abnormality in a certain mRNA abundance does not unequivocally indicate abnormality of the gene bearing its complimentary code. Second, mRNA copy numbers in cytoplasm are regulated by a number of epigenetic factors among which the post-transcriptional mRNA stability is of primary importance. Abnormal concentration of certain mRNA may result from deviant mRNA stability, thus mimicking, but having nothing to do with, presumed abnormality in transcription rates of corresponding genes. An instrument built upon so poorly understood biochemical basis can hardly serve as a reliable tool in the delicate task of diagnosis of human disease in clinical settings.