The trend was observed in both non-neoplastic versus adenoma and non-neoplastic versus thing cancer comparisons. On the other hand, comparison of normal and colitis specimens showed approximately equal numbers of genes with higher (60) and lower (73) expression levels between phenotypes. Between adenoma and cancers, however, there were considerably more genes up-regulated (145) in cancer versus down regulated (43). A summary of differential expression change by phenotype is shown in Table 1 and a list of validated genes up- and down-regulated in cancers compared to adenomas are shown in Tables S1 and S2, respectively. Probesets that revealed differential expression between neoplastic (adenomas and cancers) and non-neoplastic tissues (normal and colitis) were mapped to putative gene symbols using the most recent microarray annotation files.

108 probesets elevated in neoplastic tissues by at least two-fold were mapped to 97 gene symbols and 338 decreased probesets were mapped to 264 gene symbols (Table S3). Phenotype-specific genes The sets of differentially expressed genes were further analysed using a new analysis method developed by us to predict transcripts that may be expressed in one phenotype (e.g. neoplasia) but not in another (e.g. healthy controls). By applying this methodology, 23 probeset targets were identified as putative candidates for neoplastic-specific gene expression, i.e. hypothetically switched-on in neoplastic tissues but switched-off in non-neoplastic controls. In addition, 35 genes were identified as candidates for expression in non-neoplastic tissues only, i.

e. switched-on in non-neoplastic tissues but switched-off in neoplastic tissues. An example of a probeset exhibiting a prototypical neoplastic-specific response pattern is shown in Figure 2A, and the complete list of probesets corresponding to hypothetically switched-on and switched-off genes is shown in Tables S4 and S5, respectively. Figure 2 NFE2L3 �C prototypical ��switched ON�� gene in colorectal neoplastic tissue relative to non-neoplastic tissue specimens. Custom ��Adenoma Biomarker�� gene chip The PCA plot of the discovery data (Fig 1A) shows a strong neoplasia vs. non-neoplasia segregation involving the first two principal component axes, with the adenomas and cancer grouped together.

For the validation data (Fig 1B), the first principal component confirms that the largest source of variance across these probesets is the presence or absence of neoplasia. The second principal component, however, shows that the adenomas Cilengitide are grouped separately from the cancer specimens. Validation of candidate biomarkers for colorectal neoplasia Of the 108 probesets whose targets were hypothesized to be over-expressed by at least two fold in neoplastic discovery tissues, 103 probesets (95%) were elevated in the neoplastic validation tissues (P��0.