Contrary to one theory that the problem with protein biomarkers lies in validating candidate biomarkers, Hartwell argued that validation and even approval are not the reasons why so few reliable protein biomarkers are known. "I put the problem squarely in discovery," he said yesterday during his talk at the second annual US HUPO meeting, held here this week. "We are not discovering markers effectively."
"Basic science has not had a big impact on cancer yet," said Hartwell, who is also president and director of the
He said in particular that improving molecular diagnostics will be the "most important thing" along that path.
Hartwell pointed out that while much scientific effort has gone into finding common SNPs that correlate to risk association for cancer, he believes protein biomarkers will be more effective in risk assessment, early detection, and cancer prevention. "Proteins are going to be even more informative" than the DNA-based diagnostics scientists use today, he said.
The discovery phase for biomarkers should be ramped up and multiplexed, he said. Whereas traditional drug discovery is limited to putting just one therapeutic in a patient or model organism at a time, a single serum or tissue sample from a patient can serve as a test for thousands of biomarkers at once, said Hartwell. Multiplexing in this way could make trials much more high-throughput and accelerate protein biomarker discovery, he said.
Hartwell also highlighted a need for efficient, low-cost assays to develop and validate candidate biomarkers. ELISA tests remain very expensive and time-consuming; scientists need highly sensitive assays that are far less costly and don't require them to develop two antibodies for each protein, he said. "I'm hopeful that within the next couple of years the bottleneck of assay development will not be limiting anymore."
One approach to dealing with this problem, and others related to proteomic technology, is working in very organized teams following the model laid out by the Human Genome Project, Hartwell said. Several such international biomarker-discovery teams already exist - he listed 11 in his talk - that have chosen to tackle a specific cancer. The teams, which are required to include expertise in mass spec, informatics, antibody development, clinical work, and mouse model research, are organized under HUPO. Anyone is welcome to form a team and take part. "I think the main value of participation is going to be learning from one another what works," Hartwell said.