Biomarkers – whether they're genomic, proteomic or metabolomic – are a hot topic these days, and it’s hard to miss the fact that conference calendars are jammed with meetings on the subject. Yet, the field is still young, and researchers spend considerable time at those confabs trying to hammer out definitions on which they can all agree. Nonetheless, companies right and left are devoting considerable energy to discovering, identifying and using various sorts of biomarkers as jumping off points for new drug discovery or diagnostic/prognostic tools to aid in therapeutic intervention. The starting point for all, though, is biomarker discovery per se – and we’ll take a look at three firms that have taken very different approaches towards that end.
If R&D chiefs and clinical department heads ever needed a rock solid reason to justify their requests for additional – and substantial – funds to conduct biomarker studies on new drug candidates, they’ve certainly got one now. For Merck & Co. Inc.’s Vioxx disaster clearly demonstrates how vitally important these scientific inquiries can be.
As we all know by now, the pharma giant voluntarily withdrew its blockbuster drug Vioxx – a cox-2 inhibitor prescribed for treating arthritis and acute pain – on September 30 because new clinical trial data found that Vioxx doubled the risk of cardiovascular events, including heart attack and stroke, after 18 months of use. The three-year study, which Merck halted, was intended to show that Vioxx could prevent the recurrence of colorectal polyps in patients with a history of colorectal adenomas (benign tumors).
Vioxx, which was launched in the U.S. in 1999 and is approved in 80 countries, has been taken by more than 80 million people over the years. In 2003, worldwide sales reached $2.5 billion. That makes Vioxx one of Merck’s most important drugs, and the repercussions of withdrawing it from the market are enormous. Not only will that action shave 50 to 60 cents off the firm’s earnings per share this year, but also the withdrawal immediately cut sales revenues by 11 percent.
As well, Merck’s number one drug Zocor, for treating high cholesterol, comes off patent in early 2006, meaning the pharma’s product sales were already scheduled to take an enormous hit from generic competition. Without Vioxx sales to cushion the blow, Merck could be in real trouble. In fact, several analysts doubt it can remain as a free standing company – especially because the massive onslaught of lawsuits that are sure to follow could end up costing the pharma billions in settlements, potentially sufficient to send it into bankruptcy.
Merck said it didn’t pick up on the cardiovascular risk factor in its pivotal trials that formed the basis of Vioxx’s approval – nor did it notice any undue events during the first 18 months of the polyp trial, which it started in 2001. So could the company possibly have predicted that there would be a problem?
Well, perhaps not. But various independent researchers, including one at the FDA, had been saying for years that Vioxx use was associated with heart problems. An FDA-sponsored study that re-analyzed past data, for instance, found that Vioxx users had a greater risk of heart attacks than patients taking Pfizer Inc.’s cox-2 inhibitor Celebrex. The agency even addressed this issue, approving new labeling for Vioxx in 2002 that included information associating its use with the risk of heart attacks and stroke – as well as ulcers, bleeding and perforation of the GI tract.
Still, it took a large, placebo-controlled trial in which the drug was studied over a three-year period to convince Merck that the cardiovascular issue was sufficiently alarming to warrant an immediate recall of the drug.
Not everyone is at risk, of course: Patients taking Vioxx as a long-term therapy (greater than 18 months) seem to be most vulnerable, and even then, only about 1.5 percent (15/1000) had heart attacks at three years, compared to half that number (7.5/1000) in the control group.
But what if Merck had developed a diagnostic assay that could be used to identify those at-risk individuals? Then they could have been excluded from the trials. And practicing physicians could use a similar test to screen their patients, only prescribing the drug to those for whom it would provide a benefit.
Of course, hindsight is 20/20 – and if Merck researchers had known from the get-go that the firm’s cox-2 inhibitor carried some small but measurable long-term risk of causing cardiovascular problems, the clinical trials could have been designed to monitor this adverse event.
However, that was 10-15 years ago, when the concept of using biological markers – be they genomic, proteomic or metabolomic – to track the effects of an experimental drug on a biological system was in its infancy, if that. And the idea that one could identify a set of markers that would allow clinicians to determine whether a trial subject would respond to a particular drug was practically nonexistent. Moreover, the notion that prescribing physicians might first run a battery of diagnostic/prognostic tests on a patient before deciding exactly which drug to give that individual was far from reality.
The scene has changed dramatically, though: Today, research on biomarkers is booming as scientists seek to discover, identify and use various sorts of biomarkers as diagnostic/prognostic tools as well as jumping off points for new drug discovery.
The field is still evolving, however, and the actual use of biomarkers in all phases of drug discovery and development – including clinical trials -- is by no means commonplace, even today. But Merck’s horrific experience has proved a huge impetus to make it so.
Certainly, no pharmaceutical or biotech company in its right mind wishes to chance another nightmare like Vioxx. And it’s really inconceivable that this lesson hasn’t hit home for every pharmaceutical developer out there. If they weren’t convinced that biomarker studies were worth their while – or if they decided that these studies, which in a clinical situation require advanced trials of the drug alone as well as trials of the drug/diagnostic combination, were just too expensive – then they’ve got a crystal clear example of just how very pricey it can be to not embrace biomarkers.
“I’m already beginning to see the reverberations of Vioxx in terms of the biomarker space,” commented Stephen Naylor, a faculty member in the department of genetics and genomics at Boston University School of Medicine as well as the computational and systems biology initiative (CSBi) and division of biological engineering at MIT. “People are re-evaluating their budgets for next year and adding zeros to them. The pendulum just swung a phenomenal amount.”
“I’ve been to three protein biomarker conferences in the last several months,” Naylor said. “All the pharmaceutical companies are very excited about the prospects of what biomarkers are going to do for them.”
As it turns out, however, each of these companies may have its own particular definition of what constitutes a biomarker, for at this point, “Biomarkers are everything to everyone,” he said. And that’s despite the fact that the NIH’s Biomarkers and Surrogate Endpoint Working Group came up with a standard definition back in March 2001: A biomarker is a characteristic that can be objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes or pharmacological responses to a therapeutic intervention.
That’s still a fairly broad definition, however, so the group devised a classification system that divides biomarkers into three broad categories: Those that track disease progression over time and correlate with known clinical measures; those that detect the effect of a drug; and those that serve as surrogate endpoints.
While researchers are studying all three categories, biotech and pharma companies favor using biomarkers as drug discovery tools – not only to detect biological responses to experimental drugs but also to aid in the discovery of new targets for therapeutic intervention.
The trick, however, is coming up with the appropriate biomarkers in the first place. “Last year [at the protein biomarker meetings] some people felt the bottleneck for useful biomarkers is on the discovery side. How do you discover specific and sensitive biomarkers? Others argued that we have lots of biomarkers [already]. The problem is validating them,” Naylor explained. This issue is far from resolved, and is still a subject of debate. According to Naylor, however, discovery and validation are both big issues. “Part of the problem is that there are not a lot of biomarkers in widespread use. Discovery is exceedingly difficult. The platforms leave a lot to be desired.”
There’s one more hotly debated subject, too: Once protein biomarkers have been discovered, is a pattern of molecular signatures (spots, dots and/or peaks) sufficient (to distinguish a diseased state from a normal one, for instance), or is it necessary to actually identify those markers (by sequencing the peptides or proteins)?
“Researchers have been battling over this issue for two years,” Naylor said. But as far as he is concerned, identification is of utmost importance. “Do the biomarkers have any biological context and relevance to the biological processes you’re investigating?” Naylor queried. “You had better know what they are and have some tentative identification that makes you believe they have relevance to a biological process.”
The trouble, Naylor said, is the fact that one may be picking up artifacts of sample collection, handling, storage and/or preparation, which could then be falsely believed to be true biomarkers. “They must have biological relevance,” he said. “You must identify them.”
Nevertheless, opinions on this matter are still split, and many companies focus their energies on developing protein expression patterns that can be used to differentiate biological states.
Caprion Pharmaceuticals Inc., for instance, is using its proteomics platform to quantitatively profile the protein complements of cells. The privately held firm claims that its industrial scale technology, called CellCarta, is able to track the abundance of hundreds of thousands of peptides across hundreds of patient samples of tissue or plasma – thereby enabling the identification of new biomarkers that can be used in various ways throughout the drug discovery and development process. As well, peptides that are differentially expressed in a patient population of interest can be selected for MS-MS (tandem mass spectrometry) sequencing and identification.
The company has already struck a number of biomarker discovery deals, including with Wyeth, Merck and AstraZeneca plc. Originally, these were exploratory studies, to “find out what the technology could do,” according to Clarissa Desjardins, Caprion’s EVP of corporate development. Overall, she added, “These were incredibly successful.”
Under the deal with Wyeth, for instance, Caprion used a preclinical model of inflammation to profile plasma changes in response to the administration of Wyeth compounds. In these experiments, Wyeth wanted to determine “which drug was better; which one it should carry forward,” she said. “In animals, they were both the same.”
In general, Caprion’s proteomics technology has consistently found a “unique signature of disease” in the plasma proteome of five different diseases, Desjardins explained. “We find a set of proteins that is repeatedly and significantly different in these diseases. The data are robust in each case.”
“The efficacy of a drug is correlated with the extent to which it can take disease-related proteins down to normal levels,” she continued. For instance, Caprion has tested a number of anti-inflammatory drugs in its system: Two of those, which are on the market but not especially effective, “only reverted disease-related proteins by 30 percent,” Desjardins said. But others, which are known to be highly efficacious in humans, reverted the disease-related proteins by 80 percent. Thus, Caprion’s technology could be used by companies wishing to compare their drug candidates to those of their competitors.
The firm’s technology platform can also be used to distinguish subjects that respond to a particular drug from those that don’t, she explained. “In animal models, the proteins of responders look like the controls, and the non-responders look like those with the disease.”
But will it work in humans, as well? “We’re looking at studying this in clinicals,” Desjardins explained. And Caprion’s got several deals in the works. “In one case, the drug is already on the market. In another, the drug is undergoing regulatory review.” One always has to be on the lookout for off-target drug effects and toxicities, though, since “some drugs affect serum proteins on their own, proteins that have nothing to do with disease.”
None of these analyses requires the actual identification of the proteins involved – an area that’s trickier than you might think. “When it gets down to individual proteins, we don’t even know very much,” she said. For instance, if one finds six proteins that are up-regulated in a particular disease state and six that are down-regulated, “how do you determine which ones are specific for that disease?” she queried. “We have to eliminate confounding factors, phenomena that are coincident with the disease.” And that requires “a hard-core experimental design to narrow the choices to a specific marker.”
So, Caprion concentrates on sequencing only those proteins that are “consistently different” in normal and disease states. “We’re now able to identify 80 percent of those proteins,” she said.
While Caprion researchers rely on the differential display of peptides from plasma samples to pick out disease-relevant biomarker candidates, scientists at Aclara Biosciences Inc. focus on understanding the protein-protein interactions (including receptor-ligand complexes) between signaling pathways in cells and tissues – pathways that, once activated or inhibited, can disturb the normal cell cycle and result in a disease state, such as cancer. Indeed, the very fact that two proteins have interacted indicates that a pathway’s undergoing a change – and this event can be exploited as an early marker of cancer activity, for instance.
The new targeted cancer therapies work specifically by blocking these specific protein-protein interactions. Take Genentech Inc.’s monoclonal antibody-based breast cancer therapy Herceptin: This product targets human epidermal growth factor receptor 2 (HER-2/ErbB2), which is over-expressed in about 25 percent of all patients with breast cancer. If a patient’s cells express another HER receptor (1,3 or 4), then the tumor will not respond to the drug.
Aclara’s proteomics technology, the eTag System for pathway analysis, has enabled the firm to pick up new cancer biomarkers as well as potential drug targets. In this system, researchers add to their sample tag receptors, which are small fluorescent markers covalently linked to recognition elements (e.g., antibodies or peptides) that bind specifically to the target of interest. Next, they add a “molecular scissors” reagent, which is a light activated cleavage agent that is covalently linked to a separate recognition element – but one that also binds to the target.
When the mixture is illuminated with UV light, the scissors cause the release of tagged molecules that are bound to the same protein or protein complex (i.e., they are in very close proximity to one another). Thus, “these antibodies are both attached to the protein we’re looking for,” explained Thomas Klopack, Aclara’s CEO. “They can be the same or different antibodies, but we usually use different ones so they will attach to different epitopes on the same protein.” As well, it’s possible to “look at multiple proteins in the same biological sample,” he said.
“We modify the fluorescein molecule by adding a tail that changes the charge/mass ratio so we can separate [the released entities] on capillary electrophoresis,” he continued. The electrophoresis results in clearly defined peaks, which allows the company to identify and quantify individual proteins and dimers.
It’s also possible to hone in on activated proteins, those that are phosphorylated, for instance, Klopack said. ”Now we attach three antibodies: the scissors, the standard eTag, and a different eTag that attaches to phosphorylation sites. We will only get a read-out on phosphorylation on a protein that has the molecular scissors.”
The company has focused its energies in the cancer biomarkers field, not only internally but also through a number of collaborations. Most recently, it signed a deal to evaluate the usefulness of its technology in selecting patients for certain of GlaxoSmithKline plc’s targeted cancer therapies. GSK will provide drug-treated biological samples, which Aclara will test in its assays. The partners will then correlate the parameters measured as biomarkers with response to the drug.
And in June 2004, Aclara joined forces with the Tokyo Metropolitan Institute of Medical Sciences (Rinshoken) to conduct a breast cancer biomarker study on patient samples to validate candidate markers picked up with the eTag technology. If it works out, these markers could be used to monitor patient responsiveness to targeted cancer therapies, including Herceptin.
This alliance follows on the heels of a small feasibility study that analyzed tissue samples from patients treated with Herceptin and chemotherapy. All 13 of those subjects were HER-2 positive, but not all of them responded to the therapy (in line with the fact that only a certain percentage of HER-2 positive patients will respond to this drug). Aclara claims that its assay was able to differentiate between the two subsets to a much higher degree than other tests (such as FISH [fluorescence in situ hybridization], one of the standard pre-treatment diagnostics for Herceptin therapy).
“Only 30%-40% of the patients treated with Herceptin who are FISH-positive respond,” explained Sharat Singh, Aclara’s CTO. “Why don’t the others?” As well, patients on Herceptin therapy eventually become resistant to it. “Why do they acquire resistance?” Apparently, alternate HER pathways are being activated in these cases, he continued, and “Herceptin is not able to shut these down.”
“Patients with HER-2/HER-2 dimers all respond, but those with HER-2/HER-3 or HER1/HER-3 dimers don’t,” Singh added. Once these data are translated into a true diagnostic/prognostic assay, cancer therapy will become even more individualized. As well, since many drugs have relatively modest effects but work on different pathways, in the future it will be possible to give cancer patients a drug cocktail to manage their disease (in a manner similar to HIV therapy).
Once it joins forces with ViroLogic Inc. – which is expected to occur later this year – Aclara’s expertise in oncology combined with ViroLogic’s skill in infectious diseases will allow the new company to develop and commercialize molecular diagnostics in two broad areas of interest. As well, ViroLogic already has a commercial lab and offers patient testing services, particularly in HIV infection, where it has reportedly performed hundreds of thousands of molecular evaluations on AIDS patients to assess their resistance to various drug treatment regimens. This built-in capability will provide a straightforward way for Aclara to market its cancer diagnostics.
It will be some time before such diagnostics, which will be used to individualize cancer treatment, are available on a routine basis. In the meanwhile, assays for the early detection of cancer and neurodegenerative diseases are sorely needed. And Power3 Medical Products Inc. is well on the way to developing them. The Woodlands, TX-based proteomics company recently forged a research agreement with Mercy Women’s Center in Oklahoma City, under which it will identify protein biomarkers for breast cancer in about 600 serum samples from normal individuals as well as patients at various stages of the disease.
In recent months, Power3 Medical has also signed a research agreement with Baylor College of Medicine to discover biomarkers for metabolic syndrome and associated disorders (including diabetes, cardiovascular disease, hypertension and stroke) in patient serum and plasma samples.
The firm has already developed a non-invasive, early detection breast cancer test, which is now undergoing clinical validation at New York University Medical Center. The NAF Test analyzes fluid from the breast milk ducts to monitor 14 biomarkers that together appear to indicate the presence of breast cancer in its earliest stages. Moreover, NuroPro, its diagnostic for neurodegenerative diseases such as Alzheimer’s and Parkinson’s that monitors a panel of nine serum proteins, is also being validated in a clinical setting – this time at Baylor through an agreement with Stanley Appel, chairman of the department of neurology, who is providing the company with samples from patients he has assessed and diagnosed.
Although Power3 Medical holds the details close to the vest, its scientific focus stems from a discovery made nearly 30 years ago by CSO Ira Goldknopf. Then a researcher at Baylor, Goldknopf discovered a Y-shaped protein (protein A24) that was part of the DNA-protein complex that controls gene activity. This Y-shaped protein actually turned out to be composed of two proteins, histone H2A and unbiquitin. Goldknopf’s early discovery that ubiquitin is conjugated to other proteins paved the way for two Israeli scientists, Aaron Ciechanover and Avram Hershko, to determine that ubiquitin’s role is to tag damaged or mutated proteins for intracellular destruction by proteasomes. Hershko and Ciechanover, along with Irwin Rose from UC Irvine, were awarded the 2004 Nobel Prize in Chemistry for this discovery. (The Swedish Academy of Sciences also acknowledged Goldknopf’s early work when awarding the Prize earlier this month.)
Power3 Medical’s NAF test, at least, ties certain very early changes in the ubiquitin system with those in the pathway that leads to the development of breast cancer. According to Goldknopf, the company has identified 13 biomarkers involved in the ubiquitin system. “These are very good for testing early stage breast cancer. They are highly sensitive.” In fact, he continued, “We’ve detected these proteins differentially expressed in women who are negative for breast cancer but who have a family history. It’s a source of genetic instability that puts women at risk.”
Obviously, there are nearly as many ways to discover new biomarkers as there are companies or research groups willing to take up the challenge – and we’ve only covered three approaches here.
But biomarker discovery per se is only the beginning. If biomarkers are going to be used to identify therapeutic intervention points in metabolic pathways, that’s one thing. If they’re going to be used to follow disease progression, that’s another. And if they’re going to be incorporated into assays that will assist clinicians in selecting the appropriate subjects for trials, that’s yet another. Moreover, at some point in the future, physicians will be able to run a battery of tests on each patient before prescribing any medication – just to make sure that the drug will actually provide some benefit.
Where do we go from here? According to Boston University’s Naylor, “There are two schools of thought for the future of biomarkers. There is the new age group that thinks biomarkers will revolutionize the way things are done and reveal the complexity of biological processes. Then there are the wizened, old age guys who believe that biomarkers are important as a set of tools that add considerably to the kit we need to unravel that complexity.” That kit already contains measurements “we’ve been doing for 10-20 years,” he said.
Would biomarkers have been able to save Vioxx, if Merck researchers had known 15 years ago what we know now? If Merck had had an active biomarker discovery program to monitor the safety and efficacy of its cox-2 inhibitor (and, according to Naylor, these markers don’t even exist today), would it have been able to catch the sort of adverse event that reared its head in the long-term dosage trial?
“Possibly,” Naylor said. “If we could go back in time they might have been able to crystallize the issues.” But that’s assuming that the scientists knew what they were looking for. Otherwise, it’s still a matter of prediction – and merely identifying a set of protein biomarkers (even if those proteins have been sequenced) that track adverse cardiovascular complications, for instance, is certainly not sufficient in and of itself to establish a cause/effect relationship.
However, there were physiological data available about 4-5 years ago that cox-2 inhibitors “had significant issues,” Naylor said. “The Vioxx situation could have been avoided without the use of biomarkers.”
“Biomarkers are not a magic bullet,” he stressed. “Once we get through this first phase of defining biomarkers, they are going to be of value… but understanding what they mean is a different story and a huge issue.”
Once that information is known, it must still be put into context and integrated with the existing body of knowledge about human health and disease. “We need knowledge management and knowledge assembly to put biomarkers into a biological context,” Naylor concluded.