Recent discoveries have generated excitement about the ability of tumor somatic mutation profile to predict a patient response to pharmaceuticals . In order to fully exploit the power of somatic mutation profiling, large scale experiments that can detect somatic mutations in 1000s of exons in 100s of heterogeneous clinical tumor samples are required. We have developed a technology called Mismatch Repair Detection (MRD) based on the bacterial mismatch repair system for detecting of variants . We have engineered a specific bacterial strain (Mutation Sorter) that can sort 1000s of DNA amplicons in parallel into two pools: those carrying variations and those that do not. The identification of the fragment content of the two pools can be done using generic tag arrays allowing highly multiplexed variation detection to be performed. We have implemented the use of MRD for the identification of somatic mutations in a set of kinase genes in matched tumor and normal sample pairs. As a positive control we have identified two somatic mutations in a pair of samples known to carry these mutations. In addition we have identified 5 additional somatic mutations in 4 kinase genes and 3 mutations in TP53. Two of these mutations were present in only ∼25% in the tumor sample cells. The scalability, sensitivity, and accuracy of the MRD technology enable comprehensive scans of large gene sets to be accomplished in clinical tumor samples for the identification of mutations that correlate with cancer patient prognosis and pharmaceutical response. We will present the application of this technique to the analysis of 500 exons in over 100 tumor samples and their matched controls.

[Proc Amer Assoc Cancer Res, Volume 46, 2005]