Discrimination of Drug Sensitivity of Cancer Using cDNA Microarray and Multivariate Statistical Analysis

Practical diagnosis of complex diseases, e.g. cancer, requires high throughput profiling of gene expression in tissues. Currently, the most promising technology for it is microarrays [1]. In our method, each cDNA from a cDNA library is spotted (target) at one segment on glass-plates (4108*double/plate), with which mixture of cDNAs from normal cell mRNAs and cancer cell mRNAs (each is labeled with different fluorescence dyes; probes) are simultaneously hybridized; it finally identifies the amount-ratio of each mRNA from cancer cells versus normal cells. Such microarrays provide effective information for discriminating drug-sensitive cancers from others. We propose the discriminating method using cDNA microarrays and multivariate statistical analysis, hypothesizing that the sensitivity can be detected by the mRNA expression pattern. It normalizes data, clusters and selects representative genes for the discrimination, and finally discriminates drug-sensitive cancer from non-sensitive ones using the quantification theory.