An open letter to cancer researchers.

OVER THE PAST TWO YEARS, PANEL DISCUSsions have been held to propose recommendations to the National Cancer Institute (NCI) regarding critical initiatives in cancer biology. One outcome was a proposal for a new initiative, the Human Cancer Genome Project (HCGP) (1). This program involves systematically analyzing genomic alterations in large numbers of human tumors to determine both common genetic and epigenetic alterations and to identify changes that characterize different tumor subtypes. The price tag would be $1,500,000,000 over 10 years, the equivalent of 1000 R01 grants. We have three questions concerning the project: (i) Would the project, as proposed, achieve its goals? (ii) Would it impact ongoing or future funding for investigatorinitiated cancer research? (iii) Is this the best application of funds toward the objective of hastening the discovery of cures for cancer? A major goal of the HCGP is the identif ication of new cancer genes through genome resequencing, specifically to find mutations that occur with 5% or greater frequency across a broad range of human tumors. The implication is that such mutations would lead to new therapeutic targets. In support of this strategy, much has been said about the recent identification of activating mutations in the EGF receptor, EGFR, in lung cancer and the ability of patients with such mutations to respond dramatically to certain kinase inhibitors, such as Iressa. Mutations in EGFR exist in approximately 7% of small cell lung cancer patients, and over 50% of patients carrying mutations are responsive to Iressa, but as yet there is no increase in survival in this population. This partial success is being used as an example of the information that will emerge from the HCGP, implying that identification of many new mutations would lead to rapid cures. However, the vast majority of mutations identif ied by sequencing tumors would be loss-of-function mutations, not gain-of-function mutations as in EGFR, and thus would not be candidates for drug inhibition. Additionally, much of the excitement surrounding the identif ication of EGFR mutations was due to their ability to predict responses of lung cancers to a preexisting drug, a situation unlikely to exist for mutations in new genes. A pilot project that provides clues to the potential information yield from the sequencing component of the HCGP was recently published by Stephens et al. (2). Out of 72 breast tumors and 9 cell lines, only six kinases out of 518 sequenced had two mutations resulting in amino acid changes. Thus, mutations identified in the kinases fall below the 5% cut-off level for significance proposed for the HCGP, an already low bar. Similar results were obtained for lung and testicular cancer (3, 4). These studies addressed a family of key signaling molecules. As such, alterations in these proteins might be expected to affect cell physiology in a sufficiently broad sense to contribute to tumor development. They are also among the most easily “drugable” proteins and are a family about which we have extensive knowledge. If we cannot find interpretable