A comparison between DASL and Affymetrix on probing the whole-transcriptome

Whole-transcriptome microarray analysis has become a popular strategy to study gene-expression in cancer, exploiting the largely available formalin-fixed paraffin-embedded (FFPE) sample resources. However, there have been relatively few comparative studies evaluating the performance of the different gene-expression array platforms. We compared two commonly used whole-transcriptome microarray platforms: Illumina human whole genome cDNA-mediated annealing, selection extension and ligation (DASL) beadchip and Affymetrix U133 Plus2 GeneChip (Affymetrix). Gene expression data based on both platforms were collected on the same total RNA extracted from FFPE tissue samples of 221 advanced breast cancer patients. Correlations between two platforms were assessed using Pearson and Spearman correlation coefficients (CCs). For both platforms we also assessed coefficient of variation, which measures relative dispersion. Finally, we compared the applicability of DASL and Affymetrix for classification of breast cancer molecular subtypes using the PAM50 classifiers. Overall, there was a statistically significant, positive gene- and patient-wise correlation between the two platforms, with stronger relationship for patient-wise CC. The relative dispersion was smaller in DASL compared to Affymetrix. The consistency in subtype classification for both microarray platforms was weak (63%). We observed weak, yet positive correlation between two platforms and different magnitudes of correlations were observed according to the metrics used.