As the saying goes, all good things must come to an end. And for pharma, that means banking on the one-size-fits-all model of drug development. Companion diagnostics, a relatively new concept that's hitting the frontlines of drug discovery and development, is promising to change the way drugs are made and marketed.
With more and more biomarkers being discovered and validated, the field of companion diagnostics has more than a leg to stand on. Using a companion test — whether it looks for genetic, proteomic, or gene expression markers — to predict whether a drug will work in someone or what kind of dose that person should take is becoming more common and, according to experts, will eventually become the norm. "It's a new field, and it's growing," says Peter Tolias, executive director of the Institute of Genomic Medicine at the UMDNJ-New Jersey Medical School.
In the last five years most of the major pharmaceutical companies have adopted new programs to deal with companion diagnostic products. "In the olden days they never used to worry about who their drug was going to be functioning in, or who's going to have an adverse effect and to actually support it with a clinical product in the marketplace," Tolias says. "The tide has changed now and pharmaceutical companies are very interested in moving products onto [the] market where they can target safer and more effective products with companion diagnostics."
Recently, the field has really heated up, and that's due in part to increased awareness among pharma that in order to be more effective and save money, drugs will have to be more personalized. "The very important trend that's emerged is the realization by the pharmaceutical industry that whether they like it or not, this is just how things are going to be, and once they realize that, that should be a really big impetus in terms of moving things forward," says Stephen Little, CEO of UK-based pharmacogenomics company DxS.
While the anticipation in the field is that there will be a marked increase in the number of companion diagnostic products, it's not going to happen overnight, says Tolias. "It's going to take probably a generation for us to see a transformative change in the way that drugs are used, but clearly it's already happening and there are tangible examples on the market today," he adds.
The Personalized Medicine Coalition, a nonprofit advocacy group, reports that there are currently about 40 drugs in the US that have companion diagnostic tests associated with them — whether that means as a requirement to their being prescribed, a recommendation for use, or label information that lists genetic susceptibility relating to efficacy or dose.
While a number of pharma companies have bought in, including National Pharmaceutical Council members Abbott, Pfizer, Bristol-Myers Squibb, Boehringer Ingelheim, and AstraZeneca, among others, it's noticeably the diagnostics industry that has benefited financially. According to the PMC, the 200 companies in the diagnostics space reported $30 billion in sales in 2008. There are 65 publicly traded imaging diagnostic companies, and venture capital invested in 57 molecular diagnostic companies last year totaled $467 million.
The first test to be associated with a drug debuted in 1998. Diagnostic company Dako launched Hercep-Test, an immunohistochemistry assay used to identify patients with HER2-positive metastatic breast cancer, when it was discovered that patients with HER2 amplification responded better to Genentech's breast cancer therapy, Herceptin. In August of this year, Genentech and Dako agreed to submit for FDA approval both HercepTest and Dako's HER2 FISH pharmDx test for Herceptin in the treatment of advanced HER2-positive stomach cancer.
One leader in the field is DxS, which has partnered with some of the biggest pharma companies to co-develop companion tests. In April 2007, the company launched TheraScreen, the first CE-marked diagnostic for detecting mutations in the epidermal growth factor receptor gene. The kit is used to select lung cancer patients who can be treated with tyrosine kinase inhibitors, since recent studies have shown that some patients with non-small cell lung cancer who have mutations in this gene respond better to tyrosine kinase inhibitors such as Roche/OSI Pharmaceuticals' Tarceva and AstraZeneca's Iressa. In 2007, DxS launched its K-RAS mutation detection kit, and in 2008 the company partnered with Amgen to develop a K-RAS companion diagnostic that can be used to predict whether a patient with metastatic colorectal cancer will respond to Amgen's drug Vectibix. The K-RAS test is also used to predict response to ImClone System's Erbitux, which only works for patients whose tumors are not mutated.
"The uptake of [the K-RAS mutation diagnostic] has been really fast, and I'm sure part of the reason for that is that we don't just have the diagnostic industry promoting the benefits of the assay, we have the pharmaceutical industry promoting the benefits of the assay as well," says Little at DxS. Since the whole point of these tests is to identify patients who will do well, "you end up with better treatment for less money," he adds. "It's a very alluring prospect."
Other notable collaborations are in the works, too. In July, GlaxoSmithKline and Abbott announced that they would develop a companion diagnostic for GSK's investigational MAGE-A3 immunotherapy. In October 2008, Merck signed a two-year exclusive licensing agreement with Celera that gave the pharma access to up to 10 cancer targets for the development of RNAi-based therapeutics. Effectively, Celera can develop companion diagnostics for any therapeutics that Merck develops out of the licensed targets. In 2007, Merck inked a research collaboration agreement with Asuragen to develop a gene-expression-based companion diagnostic for use in Merck's clinical trials for an investigational cancer treatment.
Use in clinical trials is one of the reasons pharmas like to partner with diagnostic companies. According to Carol Berry, head of the pharmaco-genomic services division at Asuragen, "Our clients, which are pharmaceutical companies, mostly are looking at how can we develop assays — in other words, gene signatures — and then use those gene signatures in various phases of their clinical trials." Indeed, most pharma giants look to companies like Asuragen, for instance, toward the end of phase 2 or 3 trials when they are considering FDA submission. "The regulatory requirements drive them to solicit outside companies like ours that have the expertise and the know-how and the manufacturing facilities to be able to produce the diagnostic," Berry says.
While most available companion diagnostics — whether molecular tests like those for the EGFR mutation or in vitro diagnostic tests like Oncotype DX's test that looks for gene expression differences to measure susceptibility to breast cancer reoccurrence — have been developed separately from the drug they're used with, the trend now is toward developing them together. And as more clinical trials include early biomarker tests for toxicity or efficacy screening purposes, it's easy to springboard these into companion tests, says Tolias.
"It's pretty straightforward for us to design and develop a good quality assay, but to actually link that assay to drug response needs a clinical trial, and that needs the pharmaceutical industry's buy-in," says DxS's Little. While that's happened over the last several years, the bigger challenge is getting pharma to start planning co-development agreements earlier in the clinical trials process. "I think the biggest challenge is often one of lack of foresight," Little adds. "Oftentimes we're involved with a drug company who requires a companion diagnostic, but they need it straightaway and they haven't really given us time to develop that. What would be ideal would be if those co-development programs were initiated earlier."
From the viewpoint of PMC, whose membership comprises a number of different organizations from diagnostics firms to pharma companies, there's a more complex set of problems. "The significance here is that the companion diagnostic can facilitate the approval of a drug because it works in a large percentage of people," says Director Ed Abrahams. "[Pharma has] understood for quite some time that they need to enrich the pool — so that they understand why a drug works and for whom it works and so that they can pull the plug early if they're seeing it's not working. … But now they understand, to get something onto the market, it can be useful to have the diagnostic joined to the drug." But working together is not as simple as it sounds. Abrahams says one of the big hurdles to getting personalized medicine tests up and running is that of divergent business models, which do "not allow for easy partnerships."
A looming incentive for pharma companies to partner with a diagnostic manufacturer is that submitting a diagnostic with a drug might increase the therapeutic's chances of being approved. In the case of Novartis' failed drug, Prexige, there have been whispers about the pharma resubmitting the drug to FDA for approval, but this time with a companion diagnostic test. If submitted and approved, this would be the first time a drug has been resuscitated by a companion diagnostic.
Another big hurdle, according to Abrahams, involves regulatory issues. While FDA issued a draft Drug Test Codevelopment Concept Paper in early 2005 to lay out the initial guidelines for the regulatory process for drugs with companion tests, the document hasn't been improved upon much since then. In 2002, the FDA created the Office of Combination Products, which addresses companion devices and tests that don't fit into the agency's established pipelines. "Companion diagnostics don't really fit into the current FDA paradigm of regulating diagnostic kits," says PMC's Amy Miller, public policy expert. The FDA has indicated that it may turn the concept paper into an actual guidance for industry, and the PMC is taking responsibility for collecting suggestions from the community on how to update it. They plan to publish those suggestions in late October, says Miller. However, she adds, "It's really unclear how these products are going to be regulated."
Felix Frueh, who used to lead the genomics review group at FDA's center for drugs and now heads up research and development for personalized medicine at Medco, is on the board of PMC and has helped compile the comments. "For one, the co-development draft guidance is still only talked about, but nothing has been written," Frueh says. So far, he adds, many feel the current concept paper is too technical and doesn't address the regulatory strategies that one could actually use in the co-development of a drug and companion test. "The last I heard is that the comments are going to be written up more or less as the white paper itself," Frueh says, adding that he thinks FDA should focus the paper on teasing out the clinical and strategic issues involved in the process. "But I don't know what FDA is currently thinking about — [whether they are] going to write a guidance, how that guidance is going to look, and what it's going to say. … I know they want to do it, I just don't know whether it's going to happen or not."
A significant incentive for drug makers to co-develop a diagnostic is not only for preventing drugs that will ultimately fail, but also because FDA approval might come more easily. "The approval of the drug is potentially more imminent with a diagnostic because they can show it's working," Asuragen's Berry says. However, matching up timelines and understanding the separate regulatory processes between drug manufacturers and diagnostic companies could prove daunting. "All of us are navigating through these various organizations and entities at the FDA and trying to understand what is the best way to develop a diagnostic test that could go with a drug," Berry says.
That's not to say that FDA hasn't made inroads. Changing drug labels to include various types of pharmacogenetic information is becoming more prevalent. Of the 40 or so drugs that are associated with diagnostic tests, FDA requires a companion diagnostic test to be used for only about five. Some drugs have updated labels that include recommendations to use a test — warfarin being an example — while the majority only have updated information about the possible genetic link to side effects and optimal dose. Peter Tolias says, "You're going to see this trend increasing." In fact, in July, FDA changed the labeling of colon cancer drugs Erbitux and Vectibix to be prescribed only to people with non-mutated forms of the K-RAS gene.
It's unclear, however, how useful updating labels with recommended or informational items will be. While warfarin labeling now includes FDA-recommended genotyping for mutations in two genes that cause increased susceptibility to bleeding, the label doesn't require it. Many industry insiders have said that stronger labeling is needed to make sure the tests are actually used to guide dosage when prescribing warfarin.
Down the line
While co-development strategies and regulatory issues are big wrinkles to be ironed out, what eventually matters is whether or not doctors use the tests. And though development has taken off in terms of proving certain tests' efficacy in clinical trials, proving real-world success isn't as cut and dry. Frueh's work at Medco involves this sort of outcomes research, for which the company just finished enrollment for two clinical trials. Research "focuses on establishing the clinical effectiveness for the use of biomarkers to optimize drug therapy," Frueh says. "Our focus is the translation of these tools into clinical practice," he adds. "We want to know how they're going to perform, not just in one or two highly skilled clinical settings," but in the broader sense. This will be especially useful as more insurance companies start covering the tests — it's hard to determine whether or not a company should cover a test without a true measure of its prevention of hospitalization, he adds.
Moreover, though uptake for required companion diagnostics has been good, most doctors still don't know enough about the field to feel comfortable ordering recommended tests. Medco recently conducted a nationwide physician survey that reached about 400,000 doctors. While 98 percent agreed that genetics is important for drug therapy, only about 10 percent said they felt adequately informed about it. Additionally, about 12 percent said that they've been ordering a genetic test over the last six months and about 25 percent indicated that they're planning to order genetic test. "So you see there is tremendous growth potential," Frueh says. "It really all points to a lack of education, a lack of penetration with respect to getting the information about these tests."
Little at DxS predicts a move from "cytotoxics toward targeted therapies" when it comes to cancer drug development, while Tolias see more protein markers being used. Now, it's mostly genetic markers — tests that look for SNPs or some form of structural variation — but proteins could take the field much further. "What you ultimately want is not necessarily just something that predicts risk but something that's causative, and a lot of these markers are risk predictors, but they don't actually tell you that you're sick at that time," says Tolias.