In the middle of the last decade, the search for cancer biomarkers shifted from a type of gold rush to an elusive quest. The potential payoff is substantial: if oncologists had molecules that could help them find and treat a cancer at an early stage, they expect that they could boost survival. In addition, biomarkers could profile tumors, helping doctors and patients make informed decisions about treatment strategies.
Initially, as the human genome was published, analyzing samples using mass spectrometry looked like a potential gold mine for biomarker discovery. But even after the publication of a number of papers that described biomarker candidates, no new cancer biomarkers have entered the clinic. About five years ago, researchers realized that mining the proteome for cancer biomarkers was much more complicated than they'd anticipated. But improvements in methodology—along with increasing sensitivity in mass spectrometry—are renewing optimism that proteomics can still help uncover new cancer biomarkers.
A naive startCancer biomarker research is founded on the theory that a s a solid tumor grows, it will shed unique proteins into surrounding tissues and into the bloodstream of a patient, says Michael Gillette, a research fellow at the Broad Institute in Cambridge, MA. A decade ago, researchers were optimistic that surface-enhanced laser desorption/ionization (SELDI) and other mass spectrometry–based approaches could quickly distinguish between the serum protein profiles of healthy patients and those with cancer, and use the resulting patterns in diagnostics.
“There turned out to be an enormous number of problems with doing this: everything from the technology to bad experimental design to bad biostatistics and informatics,” says Daniel Liebler, professor of biochemistry and director of proteomics at Vanderbilt University School of Medicine in Nashville, TN. But from those missteps also came an important lesson, he says: researchers needed to distinguish between the platforms that discover new biomarkers and the ones that do targeted analysis. “It was just a mismatch of the technology to the objective.”
Tools to mine the proteome
Even with the challenges of searching for biomarkers, researchers and funding agencies are still pursuing the promise of these techniques. The National Cancer Institute's Clinical Proteomic Technology Assessment for Cancer (CPTAC) network is one such multidisciplinary collaboration that is refining methods and looking for new cancer biomarkers.
It turns out that proteins that are likely to be biomarkers are also likely to be present in low concentrations relative to the host of other proteins within the blood. Protein concentrations within blood cover a range of up to 12 orders of magnitude, and common proteins such as albumin will dwarf rare proteins. Even if abundant common proteins are removed from samples, undiscovered cancer biomarkers likely lurk among these uncommon proteins at concentrations of no more than 1ng/mL. And those concentrations hover near the limits of detection for even the most sensitive mass spectrometers.
One of the problems that plagued earlier biomarker studies was the variability and lack of reproducibility of the data. Because of the large number of peptides within a single sample, a single run of a complex mixture within a mass spectrometer does not capture all of the peptides, says Jan Schnitzer, director of the Proteogenomics Research Institute for Systems Medicine in San Diego, CA. Capturing the full complexity of a single sample can require several runs. In addition, biomarker studies have contained unanticipated biases, due to how tissue samples were handled or how the study populations were selected, says Richard Smith of Pacific Northwest National Laboratory in Richland, WA. But researchers are making progress in this area. Liebler and his CPTAC colleagues have done several studies to help standardize proteomics methodologies. Such studies have compared samples that have been stored using different methods and assessed the sources of variability looking for ways to standardize MS-MS measurements across different laboratories.
Researchers are also using a number of tools to dampen some of the biological noise to enable mass spectrometry analysis. Instead of attempting to do shotgun discovery in serum—one of the most complicated mixtures within the body—some groups have moved their initial discovery work to proximal fluids relevant to a cancer of interest, such as cyst fluid for ovarian cancer or urine for bladder cancer. The idea is that if we could get closer to the location of the tumor, Gillette says, we might be able to enrich populations of proteins that are ultimately diluted and circulating in the bloodstream. As promising candidates emerge from those samples, researchers can then see if they can be detected in blood.
