Expression profiling of archival tissues for long-term health studies

Over 20 million archival tissue samples are stored annually in the United States as formalin-fixed, paraffin-embedded (FFPE) tissue blocks, but only recently has whole-genome expression profiling from these samples become technically feasible. Here, we introduce novel general methods for assessing, summarizing, and visualizing expression data quality from archival samples. We validated these methods in technical study of 144 clinical breast cancer and autopsy samples and in an overview of all current publicly available FFPE whole-genome expression data. We additionally performed a case study incorporating over 1,000 colorectal cancer (CRC) samples collected from patients across the United States over a period of more than 25 years, integrating clinicopathological information, tumor molecular data, and archival tissue gene expression on an unparalleled scale. Both large-scale clinical studies presented a much greater range of data quality than previous smaller studies, emphasizing the need for rigorous quality control in translational applications of archival tissue gene expression profiling. This series includes 1,003 FFPE-preserved CRC tumors, assayed by Illumina HumanRef-v3 WG-DASL microarray.