In the late 1800s, Samuel Siegfried Karl Ritter von Basch invented the sphygmomanometer
the analog device most medical professionals still use to measure blood pressure. Thus came the detection and quantitation of a significant, early biomarker. Over the last two centuries, many more biomarkers emerged, including today's sophisticated molecular biomarkers, a key ingredient of modern biotech and pharma. "Right now, biomarkers lie very close to the top of the value chain," says David Persing, chief medical and technology officer at Cepheid in Sunnyvale, California. | Lynn Echan (left) and Hsin-Yao Tang (right) work on the four-dimensional protein isolation method that they helped to develop in David Speicher's laboratory at The Wistar Institute. (Image courtesy of David Speicher.) |
All companies use biomarkers. Hanno Langen, head of Roche's proteomics initiative in Basel, Switzerland, points out a wide range of possible applications, from risk assessment to diagnostics. "For example, a biomarker might tell you how a patient is likely to respond to a particular drug," he says. Moreover, biomarkers can be used early in drug development to predict toxic side-effects, and they can be used as diagnostics to assess a patient's condition.
In the October 2007 issue of Clinica Chimica Acta, for example, a team of scientists from McMaster University in Hamilton, Ontario, searched for biomarkers that would identify patients at a high risk of further problems, such as myocardial necrosis, after a visit to the emergency room with acute coronary syndrome. Such biomarkers could point out the patients who need the most aggressive treatment. These scientists assayed specimens from more than 200 patients, looking at 11 markers. One panel of potential biomarkers-interleuken-6, a version of brain natriuretic peptide (NT-proBNP), and E-selectin-picked out 60% of the patients with myocardial necrosis. As the authors conclude: "A biomarker panel analyzed early after pain onset can identify individuals at risk for both myocardial necrosis and the combined endpoint of death/myocardial infarction/heart failure." Nonetheless, the authors note that more research must be performed.
To be effective, though, biomarkers must be incorporated in a complete research program. As Persing says, biomarkers form the tip of a pyramid based on technologies needed to detect the biomarkers and the building blocks needed to use them, whether they are DNA, microRNA (miRNA), messenger RNA (mRNA), single nucleotide polymorphisms (SNPs), or the proteins encoded by these molecules. As a result, this field depends on increasingly sophisticated technology.
Searching in samples
| This image shows SGX523 (orange) binding MET (blue), a receptor tyrosine kinase that contributes to a variety of cancers. In the background, an evolutionary dendogram of the human kinome shows the complexity of finding kinases to target in fighting diseases. (Image courtesy of SGX Pharmaceuticals.) |
To identify proteins that could be biomarkers, scientists often turn to mass spectrometry (MS). Although one round of MS can reveal the components of a sample, two rounds of MS, or tandem MS, give an even more-complete view. In this so-called multiple reaction moni
In many cases, biomarkers mean proteins-such as prostate specific antigen (PSA), which has long been used as a market of prostate cancer-but many other types of molecules can be used. For example, scientists explore miRNA for biomarkers. These molecules participate in the regulation of gene expression, and they might sidestep some of the problems with using genes as biomarkers. "Thousands of genes change expression levels in cancer," says Persing. "It makes it very difficult to identify the genes that are most critical for distinguishing different types of cancer or determining the likelihood of progression." So Cepheid scientists focus on miRNA. "The data suggest that miRNAs are potentially less susceptible to individual variation," says Persing, "and they could also be more reliable as diagnostic biomarkers in general, in comparison to gene-based assays." Cepheid uses novel bioinformatic algorithms to mine the human genome for new miRNAs, and it is currently evaluating over 3,000 miRNA candidates. Among these, Cepheid scientists have identified hundreds of novel miRNAs that look like promising biomarkers for several forms of cancer, including breast, lung, and prostate.
Other targets are also gaining attention. At SGX Pharmaceuticals in San Diego, California, scientists see great promise in inhibiting MET, a receptor tyrosine kinase. It plays important roles in ordinary cell growth, proliferation, and motility, but it can also contribute to primary tumors and the onset of metastasis. "There are mutations of MET that are inherited in the germ line," says Stephen K. Burley, SGX's chief scientific officer and senior vice president, research. "This phenomenon gives rise to the hereditary form of papillary renal cell carcinoma." Even in the absence of germ line mutations, Burley adds that, "virtually all papillary renal cell tumors show evidence of MET activation."
MET also plays a role in subsets of other cancers. For example, about 20 percent of patients with non-small cell lung cancer show evidence of addiction to the MET oncoprotein (without that protein, the cancer dies). Consequently, a MET-related biomarker say, DNA, RNA, or a protein could be used as a diagnostic that determines which cancer patients should be treated with a MET inhibitor, such as SGX523, for which SGX plans to start Phase I studies in early 2008.
Trying newer technologies
| PerkinElmer's ExacTag 3.0 data viewer provides a wide range of results from protein experiments, such as the search for new biomarkers. This includes the average results for a protein, results for individual scans, and much more. (Image courtesy of PerkinElmer Life and Analytical Sciences.) |
To study biomarkers, many techniques require labels such as fluorescent dyes, but these can change a molecule's characteristics. For example, a label might make a molecule too large to interact with its typical targets. Consequently, researchers seek label-free approaches, such as surface plasmon resonance (SPR). This technique uses light to create a plasmon or group of electrons oscillating together where a metal meets a dielectric. The metal surface and the nearby molecules determine the wavelength that will generate a plasmon. In short, SPR measures the mass of particles attached to a surface, which reveals molecules binding to targets. Moreover, SPR can also reveal the kinetics of a reaction, such as a protein binding to and unbinding from a target, all in real-time. Scientists at the Institute for Systems Biology in Seattle, Washington, for example, developed an SPR-based microarray of antibodies. "We are screening serum for protein biomarkers," says Chris Lausted, senior research engineer. "The system needs very little sample, and provides high throughput." Moreover, this 800-spot microarray can be read by the institute's Plexera ProteomicProcessor. "We have found that the 100-antibody, liver-specific portion of our array detects patterns unique to drug-induced liver injury," says Lausted.
Other optical approaches could also lead to improved biomarkers. Illumina in San Diego, for instance, uses holographic codes to keep track of assays running on the surface of microbeads with its VeraCode technology. These cylindrical beads 240 microns long with a 28-micron diameter can be used to discover and validate various types of biomarkers, including DNA, RNA, and proteins. These microbeads work with many assays, including genotyping, gene expression, and protein-based assays. The holographic codes let researchers make blends of beads to run multiplexed experiments, all while keeping track of targets, the sample batch, reagents, and so on.
Other companies also offer new techniques. At the Human Proteome Organization's 2007 meeting in Seoul, South Korea, for instance, PerkinElmer Life and Analytical Sciences launched several new products for quantitative protein-expression profiling, including its ExacTag S. Peter Banks, PerkinElmer's technology and business development leader, says, "This product enriches low-abundance proteins from serum or plasma samples by depleting albumin and IgG, leading to more efficient labeling of lower-abundance proteins. We've seen up to five-fold enrichment." Such tools can be used to find new biomarkers or unravel how they relate to disease or treatment pathways.
Multiple passes
Finding the best biomarkers will probably depend on developing even more-advanced techniques. For example, biomarkers for cancer tend to be low-abundance proteins that occur with many other proteins in plasma and serum. Likewise, diseases tend to set off an acute-phase reaction, which starts lots of proteins going up and down maybe none of them valuable as biomarkers. According to David Speicher, professor and chair of the systems biology division and director of the proteomics laboratory at The Wistar Institute in Philadelphia, "If you are looking for cancer biomarkers in the high- and medium-abundance proteins, I'd argue that you are wasting your time."
To study rare proteins, Speicher's research team developed a four-dimensional approach. First, he runs a sample through three levels of protein separation: major-protein depletion, microscale solution isoelectric focusing, and then a one-dimensional SDS gel. The first pass protein depletion provides two fractions, and Speicher works with the one without the abundant proteins. Next, isoelectric focusing generates 4-6 fractions. Speicher runs each of those through an SDS gel. Last, he uses mass spectrometry to identify the isolated proteins. "The power of the method and the depth of analysis arise from a combination of the number of fractions and the different dimensions of separation," says Speicher.
Although this approach gives Speicher a chance to explore an entire sample in great depth, it takes lots of work and is not high throughput. To circumvent both of these limitations and the extensive variability of blood-protein profiles in humans, Speicher primarily uses mouse cancer models. For example, when using xenographic mice models of human cancer, Speicher uses differences in protein sequences to look for proteins shed by a human tumor in the mouse. "We've seen proteins that have not been previously reported," he says. "Now, we need to select the best candidate biomarkers and then validate them in serum or plasma from human patients and matched controls."
At Merck, scientists also make multiple passes in search of biomarkers. For example, Merck collects tissue samples from patients and analyzes them for biomarkers related to a specific kind of cell or in response to a specific clinical condition. This work involves expression profiling of mRNA, protein profiling, and genetic profiling. Merck uses the resulting biomarkers for many purposes, including testing drug efficacy early in the development process. Moreover, Merck applies molecular profiling in all of its therapeutic areas. It even extends this work through collaborations, including working with the Harvard Brain Bank to study a new target that might be related to Alzheimer's disease.
Computing new biomarkers
Part of Cepheid's search for miRNA biomarkers depends on computation. Many structures in the human genome look like they could make miRNA. To find new miRNAs, Cepheid scientists developed an algorithm that correctly identifies known miRNAs about 70-80 percent of the time, as well as thousands of new ones. Cepheid scientists have used these algorithms to predict miRNAs and then put them on a microarray. In about 30-40% of these microarray experiments, Cepheid finds evidence of up or down-regulation of these miRNAs in either normal or diseased tissue. "This doesn't mean that the remaining 60 to 70% are not expressed," Persing says. "More likely, we just haven't tested the right tissue, developmental stage, or conditions of stress for those candidates to be seen." Persing thinks that there could be thousands of miRNAs, maybe tens of thousands. In fact, Cepheid has already completed the first level of validation on about 700 miRNAs that are not in the public databases. These new miRNAs could turn into biomarkers for various diseases, either to track the progression of a disease or to track the efficacy of a treatment.
At Roche, scientists also rely on computation combined with data from high-density SNP chips. "With increasingly specific genetic data on SNPs from more individuals," says Mitchell Martin, head of Roche's molecular medicine group in Nutley, NJ, "we can map variation and look for correlations with a specific endpoint, such as likely drug response or, in some cases, potential risk behind a compound." He suspects that truly unraveling the information behind the growing volume of data will require advances in statistics that keep pace with the data-collection technologies. "The whole field of statistical genetics is just catching up in terms of methods," he says.
The trick for tomorrow, says Burley, arises in trying to turn biomarkers into diagnostics. "That's where you'll see innovation," he says. As one ongoing example, he points out the work of Josh LaBaer's team at the Harvard Institute of Proteomics. Instead of going to the trouble of expressing proteins in cells, making them ready as probes, and spotting them on microarrays, the Harvard team spots arrays with DNA and then transcription and translation grow the protein probes on the chip. Burley also likes the system at CellFree Sciences in Japan, which uses technology developed by Yaeta Endo of Ehime University. This system starts with wheat germ, which gets turned into an extract that makes proteins based on added mRNAs. These systems could increase the throughput of biomarker-based diagnostic tests.
The ongoing advances in biomarkers also lift the entire field of pharmaceuticals. Virtually all of the technologies mentioned here work for target identification and validation, as well. "Today's targets are more likely to be in pathways that are causally involved in a disease and can be modified to improve a patient," says Martin. With the best targets and biomarkers to track them, researchers can understand diseases more completely and treat them more safely and effectively.