| “Pharmacogenomics.” Now there’s a word that glitters with promise. By measuring human genetic variations and correlating them with individual responses to drugs – or by using them to predict these responses – pharmacogenomics is guaranteed to revolutionize the entire drug development process. All pharmaceutical companies – from the largest drug powerhouse to the smallest biotech firm -- acknowledge that this will occur sooner or later. Most of the major players, however, are hedging their bets: They’re stockpiling samples of DNA, blood or tissue from patients enrolled in clinical trials, but they’re not actually doing anything with them. These companies keep the samples in reserve, only to be analyzed if an adverse reaction occurs during – or after -- a trial. Many biotech firms, on the other hand, are out to show that pharmacogenomic studies can easily and affordably become an integral part of the clinical trial process – and that the results will prove indispensable in designing future studies. In fact, they’ve already proved it. |
Big pharma is gearing up for practical pharmacogenomics. Reportedly, most if not all of the majors have already started stockpiling samples of DNA, blood and/or tissue from patients enrolled in clinical trials – samples that will tell them which patients are genetically wired to respond well to an experimental therapy, as well as those who are poor responders or over-responders. These precious samples should also offer valuable clues as to whether trial enrollees are genetically predisposed to develop a certain type of cancer, for instance, or high blood pressure.
Indeed, stored specimens will yield scads of useful information – once the companies get around to actually doing the analyses. For, as it turns out, most big pharmas are not performing large-scale gene expression or genotyping studies on clinical trial participants at the moment – nor are they particularly inclined to do so. Especially because these sophisticated experiments are still relatively expensive to perform, the preferred approach is to analyze samples only if subsequent events warrant it. For instance, if a subgroup of trial participants experiences side effects while taking an experimental drug, the sponsoring company could go back to the vault and dig out only those patients’ samples for extensive testing. Or, if some patients respond positively to the treatment, while others might as well be on the placebo, then retrospective analysis could guide the design of the next clinical trial. Moreover, these samples will come in handy if and when the FDA requires genotype information as part of the drug application process.
In truth, pharmas are still more interested in identifying and dumping toxic drugs early on than they are in developing personalized medicines – though they may some day use archived patient samples to create diagnostic and prognostic tests for developed drugs.
Some say that big pharma’s hesitancy to jump head first into pharmacogenomics has impeded deal-making, although Recombinant Capital’s databases contain more than 60 biotech-big pharma alliances with a pharmacogenomics component that have been initiated since the beginning of 1998. This certainly means that not all drug-makers are so conservative. In fact, as we’ll see, some of them have already signaled a major commitment to pharmacogenomics through their alliances.
But if the others are waiting for proof of principle, and assurances that pharmacogenomic studies can become an integral part of the clinical trial process, then biotech firms are ready, willing and able to convince them. And, if the majors are holding back until the technology’s been refined, then biotechs are even now working out the best approaches for identifying appropriate biomarkers, SNPS and HAPs, for performing large-scale analyses on patient samples and for standardizing sample collection and storage. We’ll highlight the programs underway at a select group of biotech companies, providing a small yet representative sample of the broader efforts underway within the sector.
One clinical trial, in particular, has already provided extremely compelling evidence that pharmacogenomics can and does work in a real-time situation. This trial -- Genaissance Pharmaceuticals Inc.’s STRENGTH study – demonstrated that gene variations are associated with individual responses to a variety of statin medicines – including Pfizer Inc.’s Lipitor, Merck & Co. Inc.’s Zocor and Bristol-Myers Squibb Co.’s Pravachol (and, before it was recalled due to a side effect, Bayer AG’s Baycol) -- which are used to treat people with high levels of cholesterol. In this 450-patient trial, results of which were first reported last month at the 52nd annual American College of Cardiology meeting in Chicago, Genaissance used its haplotype technology to construct genetic markers that were prospectively analyzed for their association with the reduction of LDL-cholesterol. (Statins work by inhibiting an enzyme that controls the rate of cholesterol production; by slowing down cholesterol production, the liver is better able to remove LDL-cholesterol from the blood.)
According to Gualberto Ruano, vice chairman and CSO at Genaissance, “We selected the markers based on the pharmacology of the statins and the mechanism of action of cholesterol-lowering agents.” The company combined a candidate gene approach with an analysis of the cholesterol metabolic pathways (in which about 75 genes are involved).
The researchers found that a specific variant of a gene involved in cholesterol metabolism controls LDL-cholesterol response to statins. They found a statistically significant association between a haplotype marker in the gene ABCA1 (a transporter gene, related to cholesterol homeostasis and one not previously identified as affecting statin efficacy) and LDL-cholesterol reduction in treated patients. This association was found with each individual statin, too.
And because the researchers took a blood sample from each patient and made individual cell lines, from which they then extracted the DNA, the company has plenty of material with which to work in the future. “The analysis [of the trial results] is ongoing,” Ruano explained. How long it will take “depends on the questions you ask.”
The results demonstrate that “Pharmacogenomic predictions are accurate and powerful,” Ruano said. “This bodes well for the future of pharmacogenomics.” Moreover, “We should start thinking about a genetic signature or structure for a drug, and build a different kind of structure/function relationship.”
Genaissance, which originally intended to delineate genetic variation by measuring SNPs (single nucleotide polymorphisms), quickly turned to HAPs (haplotypes) instead. Researchers know that genetic variation among individuals is organized into "DNA neighborhoods," or haplotype blocks, which have remained largely intact throughout human evolution. Since the SNPs on any one haplotype block tend to be inherited together, scientists can simply tag a few variants in a block to be able to identify that block and know all its SNPs. Thus, this shortcut should reduce the number of SNPs that researchers need to study from perhaps 15 million to about 300,000 to 600,000. (One individual’s genome contains about 3 million SNPS and it’s been estimated that a population may contain as many as 15 million variations.)
And, since most drugs interact with the product of more than one gene, it’s necessary to analyze variations in multiple genes to predict or chart a response to a particular drug. Genaissance’s goal is to have a database containing the haplotype to each and every one of the genes that are related to drug response. The company’s HAP database already contains information for more than 7,000 pharmaceutically relevant genes.
The STRENGTH trial is Genaissance’s crown jewel, but the New Haven company has also conducted two other clinical studies analyzing drug responses. In these studies, too, the company claims that it has conclusively shown that HAP markers can identify patient populations with a defined drug response. “The STRENGTH study was special because of the sheer scale of the experiment. The other two internal studies were much smaller,” Ruano said.
The company’s first clinical study, intended as a proof of principle, involved examining genomic variation in the beta2-adrenergic receptor, the target of the asthma drug albuterol (GlaxoSmithKline plc’s Ventolin), to ascertain whether it could use HAP markers or SNPs to define an asthma patient population that responded to the drug. The 121 asthmatic patients enrolled in the study varied in their responses to the drug; in fact, albuterol was clinically effective in only 40 percent of those patients. The researchers found that specific pairs of HAP markers in the receptor gene were carried by those patients who exhibited a positive drug response, and poor responders carried other specific pairs of markers. This correlation was statistically significant. By contrast, the company found that no individual SNP had equivalent predictive power.
Genaissance is currently recruiting patients for its CARING study, to identify HAP markers that define patients with schizophrenia who are most likely to develop agranulocytosis (a potentially life-threatening depletion of white blood cells) if treated with the schizophrenia drug clozapine. If the company succeeds in identifying these markers, it intends to line up a partner to commercialize them – either through a stand-alone diagnostic test or a test associated with a drug. Genaissance used the same sort of strategy early this year to parlay its findings from the STRENGTH trial into a commercialization deal with Bayer HealthCare LLC, which gets exclusive rights to develop a diagnostic and non-exclusive rights to develop drugs.
And, in another deal (of which Genaissance has many), Biogen Inc. will use Genaissance’s HAP technology to study the pharmacogenomic basis of patient variability in response to Amevive, Biogen’s newly approved biological therapy for moderate-to-severe psoriasis. As part of the January 2002 alliance, Biogen also gets a license to use those HAP markers that are shown to be predictive of response to Amevive to develop diagnostic tests for use in conjunction with its product marketing efforts.
Vertex Pharmaceuticals Inc. is also incorporating pharmacogenomics into its clinical trials on VX-148, an experimental treatment for moderate-to-severe psoriasis. As part of a broad agreement signed in January 2003, Iceland’s deCODE genetics Inc. will gather and analyze pharmacogenomic data as part of the clinical trials that deCODE’s subsidiary Encode will conduct on Vertex’s compounds. The first of those (now in Phase II trials) is VX-148, a second-generation small molecule inhibitor of inosine monophosphate dehydrogenase (IMPDH), an enzyme that plays a key role in regulating immune response and the proliferation of certain types of cells, including lymphocytes. By performing pharmacogenomic analyses in parallel with clinical trial results, Vertex will obtain information about the genetic factors that affect the responses of individuals to its experimental drug. The data should serve to guide the design of the next set of trials and lead to pharmacogenomic tests that can be used in conjunction with the drug.
In general, Encode’s procedure involves genetic profiling to prospectively identify responders and non-responders. A small number of patients are given the drug of interest to establish response baselines. Blood samples from these individuals are then prepped and challenged in vitro with and without the drug to generate gene expression profiles. From there, the scientists select a panel of differentially expressed genes that most accurately predicts the clinical response to the investigative compound. (This effort is helped tremendously by the fact that deCODE already has a computerized database containing phenotypic and genotypic data on tens of thousands of individuals who have participated in the company’s genetic research into more than 50 common diseases.)
“Basically, we use genetics to look at biological pathways,” explained Kari Stefansson, deCODE’s president and CEO. “We look at which part of the population has a particular disease on the basis of pathways. We select people for a trial on this basis. This allows us to subdivide the population and match the drug and the mechanism by which the patients developed the disease.”
The genes can be further analyzed for SNPs that correlate with responsiveness to a drug, allowing the creation of a DNA-based predictive assay. In addition to these tests, which one day will be marketed in conjunction with drugs, “It’s absolutely clear that pharmacogenomics will play an important role in drug development itself, to select and stratify patients for clinical trials,” Stefansson said. “The use of pharmacogenomics in clinical trials will be the key to successful drug development in the future.”
You’d be hard put to find anyone in this crowd who’d disagree with that statement – but the trick is to convince the pharmaceutical majors that pharmacogenomics is ready to go into the clinic now. It’s a tough sell, but not as tough as it used to be: Thanks to rapid advances in the field, and proof-of-principle studies, big pharmas are increasingly willing to give it a try through collaborative efforts with biotechs.
Just ask Perlegen Sciences Inc. When it was spun out of Affymetrix Inc. in the fall of 2000, the new company’s goal was to adapt Affymetrix’ DNA scanning technology to read 50 genomes and identify the millions of genetic variations between individuals (i.e., SNPs, of which Perlegen has now identified 1.7 million) – and then to find patterns in those variations.
Well, the privately held and richly financed company completed its gargantuan whole genome-scanning task late last summer and scaled up operations so that it could sequence one genome in about 10 days. As soon as it had achieved these goals, the company started lining up big pharmas to help associate those patterns with health factors and drug responses. In rapid order, Perlegen struck deals with GlaxoSmithKline, Eli Lilly and Co., Bristol-Myers Squibb (BMS) and Pfizer. In the BMS collaboration, for instance, Perlegen will scan the entire genome of hundreds of clinical trial participants to identify markers for patient response to drugs in BMS’ portfolio. And for Pfizer, Perlegen will identify variations in DNA that are associated with cardiovascular disease – discoveries that can then be used to develop new drugs and diagnostics.
And, though pharmaceutical companies are interested in gathering information on the “inherited genetic determinants of disease as well as individuals’ response to drugs, they are more interested in the drug response,” according to Brad Margus, Perlegen’s CEO. And for that, it’s not necessary “to understand the biology.”
“The pharma supplies the patient sample plus the clinical data to go with it,” explained David Cox, Perlegen’s CSO and co-founder. With Perlegen’s high-resolution, high-speed and cost-effective technology, he said, “We can look at [the genomes of] hundreds of people” to identify the genes responsible for common diseases and how these genes affect people’s responses to drugs.
Indeed, the company’s massively parallel platform allows it to design studies with significant statistical power. For instance, a clinical trial might involve 400 control patients and another 400 who receive the experimental drug. “We look at 1.7 million SNPs in 800 individuals,” Margus said. “We repeat this multiple times to eliminate false positives and then confirm the findings in a different set of samples from a different set of people with the same condition.” Such large-scale analysis should allow researchers to “find numerous genes with a small effect,” the sorts of genes that work in concert to cause common diseases.
It’s this sort of analysis that will drive pharmacogenomics into the clinic, Margus believes. “If we can discover association data that are really useful, it will make for a much better drug. And if it really is a valuable drug, then it will get to the market.”
But there’s another way to dig out the root causes of disease – and to understand the molecular basis for patient-to-patient variations in disease presentation and response to therapy. Rather than analyze the DNA per se, SurroMed Inc. looks for markers present in blood and other biological samples. The Mountain View, CA company has developed techniques for profiling and analyzing thousands of immune cell populations, proteins and small molecules such as sugars, lipids and peptides. When integrated with clinical data, this approach should allow the company to identify the biological pathways involved in disease and therapeutic response.
The company, which was founded in 1997, is “still in biomarker discovery mode,” explained Gordon Ringold, SurroMed’s co-founder, CEO and chairman. “We’ve developed a system for discovering thousands of markers in a highly automated, computational manner… We measure as many markers as we can that correlate best with the presence or absence of disease, but we combine our technology with the best hypothesis-driven approaches,” he said.
And SurroMed has already signed on a number of partners – which will be supplying clinical samples for analysis. One of them is Biogen, which is interested in studying the response of multiple sclerosis (MS) patients to Avonex therapy as well as identifying biomarkers that can be used to develop new MS drugs and associated diagnostic tests.
Each of SurroMed’s biomarker alliances is unique when it comes to the clinical design and the types of samples to be analyzed, Ringold said. “The protocol is highly dependent on the patient population, the disease and the indication.”
The company’s also conducting its own biomarker studies in rheumatoid arthritis (RA), which Ringold says is “almost certainly not one disease, but rather five or six” separate diseases that are currently lumped together. “We have two different RA clinical studies that are not partnered. In the first, we’re enrolling RA patients already defined as RA patients as well as some control subjects who don’t have RA. We’re also trying to enroll patients with potential early signs of RA so we can identify early disease,” he continued. The trial subjects are sampled on a quarterly basis and their blood analyzed. “The things that we measure are different in the controls and the patients. We generate a list of candidate biomarkers… then analyze a subset of those that change (up or down) in the same patients over time.” These results can be correlated with disease progression in individual patients – as well as their response to therapy.
The second RA study is being conducted under an NIH contract, and involves patients, originally drug-naïve, who are then put on anti-TNF therapies, including Remicade and Enbrel. The goal is to identify biomarkers associated with patient response and non-response to these drugs. The markers will also be used to create diagnostic products.
Eventually, Ringold said, the biomarkers will be able to predict the responders and non-responders, even before they take the drug. “We see this paradigm beginning to emerge as we try to get more precise in prescribing these kinds of molecules.”
Ultimately, of course, researchers hope that a complete understanding of disease pathways will lead to actual disease prevention rather than treatment. “Our goal is to find these markers earlier and earlier in the disease process,” Ringold said. But in the early stages of a disease, the changes can be very subtle – which makes it doubly important that the biological samples SurroMed analyzes are prepared and handled properly. In fact, he said, it’s critical. “We have to adopt a systematic approach to minimize the differences in handling and processing as well as focus on the reproducibility from the collection site.”
Indeed, sample handling and processing will become an increasingly important issue in the future. Once pharmacogenomic studies become an integral part of clinical trial design, each and every trial subject across the U.S. will be donating samples – blood, tissue, cerebrospinal fluid, urine – for further analysis. Likewise, hospitals and medical centers already collect specimens (including biopsies) from untold numbers of patients. All those samples have to be stored somewhere – and if they’re not prepared, processed and documented appropriately, they’ll be worthless.
It comes as no surprise that a number of companies have sprung up to address this situation. Privately held Ardais Corp., for instance, has amassed a collection of over 140,000 research-quality tissue samples and associated clinical data representing a broad array of diseases.
Researchers can access this information – and the samples themselves – through Ardais’ online data library. It’s a service that comes at exactly the right time – as witnessed by the fact that the Lexington, MA company already has inked clinical genomics alliances with about 36 pharmaceutical and biotech companies (including AstraZeneca plc, Aventis SA, Bristol-Myers Squibb, Abgenix Inc. and CuraGen Corp.)
The samples are collected through the National Clinical Genomics Initiative, a collaboration between Ardais and four medical centers – Boston’s Beth Israel Deaconess Medical Center, Duke University Medical Center in Durham, NC, Maine Medical Center in Portland, ME and the University of Chicago – that was set up specifically to develop systematic, large-scale procedures for collecting, processing and storing research-quality clinical materials and associated information.
The entire process involves legal and ethical issues, as well as technical ones, according to Alan Buckler, Ardais’ co-founder and CSO. The company set out to address these issues, to standardize sample collection and collect clinical data and then structure everything so researchers can actually use it, he explained.
For instance, “the physical protocol [for collecting samples] is documented every step along the way. Validation of every step is very important.” The samples also undergo a pathology verification process to ensure that they are of research quality. Very importantly, informed patient consent is an integral part of the procedure. Patient samples are de-identified, as well, assuring confidentiality and protecting privacy. And, Buckler added, “Medical centers are very motivated” to participate in this initiative. “It will benefit the patients they serve.”
While Ardais specializes in tissue collection, privately held GenVault Corp. stores DNA samples. The Carlsbad, CA company has created a DNA archiving system that not only allows clinical DNA samples to be stored and retrieved but also links the DNA with relevant clinical information (which has been de-identified to protect patients’ privacy). The company’s not-so-modest mission is to build a network of archives across the country that will serve as an efficient exchange for biological samples and information.
This is “high throughput DNA archiving,” explained Mitch Eggers, the company’s president and CEO. The DNA samples are stored, retrieved and shipped in a room temperature, dry format – “a fast and inexpensive way of handling physical samples.”
When a patient sample is collected, a small amount of blood (2-4 milliliters) is deposited onto a 384-multiwell plate. Then all the cells are lysed, leaving the DNA attached directly to the well. Once the samples are dry, they can be stored and shipped easily – and don’t require biohazard treatment or special handling, he added.
GenVault, which just last month raised $10 million in a Series A venture round, is nearly finished beta-testing its system and plans to start marketing it this summer. Researchers wishing to access the system can “search through a relational database on our website and select samples. This is enabling technology: They are no longer limited by lack of access to samples.”
If a client requests a sample, “the 384-microwell plates are retrieved and a 96-well daughter plate is constructed,” with each of those 96 wells containing a different patient’s DNA. Then the DNA can be PCR’d directly or eluted off the element. Importantly, the amount of DNA in each well is enough for “tens of PCR-based experiments,” Eggers said. And, if it’s a very popular sample, there are ways to amplify the DNA from one well, making enough to fill two to four plates. “It’s an inexhaustible supply.”
“Phase IV trials are very conducive to this type of archiving,” Eggers explained. “If there’s an adverse drug reaction, the company can go back and genotype those patients.” A diagnostic test could follow, meaning that the drug would not have to be pulled off the market. Pharmas market drugs to a broad population at the beginning, he continued. They don’t segment the market until there are side effects. “You don’t restrict the market until you need to.”
But down the road, when personalized medicine becomes a reality, GenVault’s DNA archiving system will also come in handy. “A dynamic working archive will be very instrumental in the process.” Having a large relational database “will change the way that genomic research is being done,” Eggers said.
Systems such as those being developed by GenVault and Ardais will become critically important as biotech and pharmaceutical companies gear up for large-scale pharmacogenomic studies. But there are other factors that will help shape the future of pharmacogenomics.
The most important of these may be the FDA’s new agency-wide initiatives to speed the development of new medicines. Announced on January 31, 2003, this broad-ranging plan aims to improve the transparency, quality and efficiency of agency processes and agency-sponsor interactions. And the FDA has identified pharmacogenomics as one of the key areas of emerging technology – especially as it relates to developing new therapies in combination with genetic or phenotypic tests that can identify responders, customize dosages and weed out individuals who may develop side effects.
If it keeps to its predicted timeframe, the agency should be issuing a set of draft guidelines on when and how to submit pharmacogenomic information to the FDA during drug development within 18 months.
According to Genaissance’s Ruano, “There’s a tremendous amount of interest in the FDA [guidelines].” Pharmacogenomics companies are anxious to see the guidelines, because they will certainly change the way that drug applications are filed – as well as the type of information that will be required on the label of an approved therapeutic. “What is the role of data in the application? Does a therapeutic have to be coupled with a diagnostic? How is this going to appear on the label? These are three major issues,” he explained.
Other events are raising the profile of pharmacogenomics, Ruano said. For instance, the current focus on biodefense – creating new vaccines and antidotes for biological weapons -- plays right to pharmacogenomics’ strengths. “We need to develop products for biodefense and security. There’s a different time track for approval, and we will have to expedite development no matter what. We can do this with pharmacogenomics,” he added.
But, perhaps the most immediate factor that’s driving pharmacogenomics forward is this: New data coming from clinical studies are providing clear proof that pharmacogenomics works. As deCODE’s Stefansson puts it, “This is not a pipedream. This is today.” |