for primaries the median was 70% for liver metastases the median was 55%, and for the carci nomatoses 80%. The samples are taken from a research bio bank registered at the National Health Institute and the project is approved by The Norwegian Data Inspectorate according to the national legislation. TP53 mutation status DNA was e tracted from tumor tissue pieces neighboring the ones used for RNA e traction. All tumor samples were previously analyzed for TP53 mutations within e ons 5 8 by screening for aberrantly migrating PCR fragments in constant denaturing gradient gel elec trophoresis followed by identification of the specific mutations by direct sequencing. Total RNA e traction The tissues were ground in liquid nitrogen and homoge nized with a pellet pestle motor in 1ml of Trizol. 0.

2 ml of chloroform was added and the samples were vigorously shaken for 20s, and then incubated at RT for 5 min. After centrifugation at 12,000 g for 15 min, the aqueous phase was mi ed with 0. 5 ml isopropanol. The RNA was allowed to precipitate for 10 min and collected after centrifugation at 12,000 g for 10 min at 4 C. The RNA pellet was washed with 75% etha nol, collected after a brief centrifugation, air dried, and re suspended in H2O at 55 C in 10 min. The purified RNA was quantified by spectrophotometer, and the quality was evaluated by capillary electrophoresis. E pression profiling For each of the test and reference samples, 20 g total RNA was reversely transcribed using the Agilent direct label cDNA synthesis kit according to the manufacturers directions.

As a common reference for all samples, we used the Universal Human Reference RNA, containing mRNA from ten cancer cell lines. cDNA was labeled with cyanine 5 dCTP for test samples and cyanine 3 dCTP for the com mon reference, and was purified using QIAquick PCR Purification col umns. The cDNA was suspended in hybridization buffer and hybridized to Agilent Human 1A v2 22 k oligo microarrays for 17 h at 60 C according to the Agilent protocol. The slides were scanned by a laser confocal scanner. Microarray data analyses The image processing was performed with Agilent Feature E traction 7. 5. Local background subtraction and linear LOWESS normalization were per formed. Semi processed values were imported into BASE, where spots with inadequate measurements were flagged and ratios calculated.

Oligonucleotide probes with inadequate measurements in more than five of the 29 tumor samples were e cluded from the analyses. For further analyses, we used data corresponding to 18 264 unique gene bank accession numbers, represented by 16 553 unique GSK-3 gene symbols. BAMarray 2. 0 was used with default settings for detecting differentially e pressed genes between two or more groups. BAMarray uses shrinkage estimation com bined with model averaging. This provides a good balance between false rejection and false non rejections.