John Harvard's Journal
Systematic Drug Discovery
Midway through clinical trials for the experimental melanoma treatment PLX4032, researchers were convinced they had a miracle drug. Patients on PLX4032 had shown significant tumor shrinkage within weeks of beginning treatment, a radical change from the effective death sentence that is metastatic melanoma. And this was no ordinary treatment. PLX4032 was among the first apparent successes in the field of targeted therapy. The drug was directed to a specific, cancer-causing mutation present in more than half of all melanomas, and its success seemed to herald a new age of personalized medicine.
What followed was heartbreaking. The drug’s early successes were followed by the sudden emergence of resistant tumors. One after another, patients relapsed. The drug that seemed to snatch them from the jaws of death wound up delaying disease progression by only an estimated six months. PLX4032 eventually received federal approval as the drug Vemurafenib, but its results fell far short of its initial promise.
“We need to reexamine the fundamental science behind drug therapy,” says Peter Sorger, Krayer professor of systems pharmacology at Harvard Medical School (HMS) and head of the new Harvard Program in Therapeutic Science. The pathway to federal approval is littered with failed drugs, representing many years of labor and millions of dollars of investments; indeed, an estimated 70 percent to 75 percent of a successful drug’s price reflects the cost of earlier losses during development. Even as science crafts increasingly sophisticated techniques for understanding chemical action at the level of molecules, the number of drugs approved by the Food and Drug Administration (FDA) has declined from approximately 100 to about 30 per year in recent decades.
To tackle this stark reality, the new HMS program aims to use multidisciplinary approaches from systems biology (a new discipline that uses quantitative and computational methods to study emergent behaviors of biological components; see “Seeing Biological Systems Whole,” March-April 2005, page 67) to create a more rational basis for drug development. “We don’t know why most drugs work,” Sorger says. As the case of PLX4032 shows, drugs often have unpredictable side effects and remarkable variation in efficacy from one patient to another. Sorger, who has co-founded two companies himself, Merrimack Pharmaceuticals and Glencoe Software, sees research potential in the problems that plague pharmaceutical companies, regulators, and clinicians. “If you were to work closely with a pharmaceutical company,” he says, “you’d continuously find these fantastically interesting biological questions spinning out. These questions are usually shelved in industry because timelines are tight, but they come back to haunt you over and over again.”
Underlying the new initiative is the belief that drug discovery has been too focused on a reductionist approach. Historically, pharmacology has been focused on the idea of a magic bullet—a single drug for a single disease process, says Joseph Loscalzo, Hersey professor of the theory and practice of physic at HMS and chair of the department of medicine and physician-in-chief at Brigham and Women’s Hospital (BWH), who is involved with the therapeutics initiative. Typical drug-discovery methods begin with high-throughput screens that identify single molecules that interact with a particular target—a protein or signal receptor that is known to go awry in disease. Increasingly sophisticated technologies can provide a detailed understanding of how the drug and its target interact at different dosage levels across time and cellular space, but Sorger nevertheless views this knowledge as insufficient. PLX4032, for instance, was derived from this highly targeted approach—a single drug for a single cancer with a single mutation—but the cancer cells’ rapid resistance, as observed in the clinical trials, also shows how much remains to be learned. “In the case of Vemurafenib, we need a much more sophisticated understanding of the drug pathways,” he says. “Most of that massive resistance is due to bypasses, where one pathway turns off and the next turns on.”
Sorger strongly advocates the use of more mathematical and computational methods to supplement biology’s traditionally descriptive approach. When he taught at MIT, he co-founded its Computational and Systems Biology Initiative, and now his lab uses quantitative models to study the biological circuitry controlling decisions about programmed cell death, a process radically altered in cancer cells. “My interest in quantitative methods grew organically from being incredibly dissatisfied with this very anecdotal picture,” he says. The new program in therapeutics will draw on MIT’s position as a leader in computational biology, and Harvard’s in systems biology (driven by the creation of the department of systems biology at HMS in 2003). “We need a radical rethink in the way that we organize and interpret biological data,” he says. A more integrative research approach, based upon predictive models of cellular networks, will help explain and predict drug side effects and interactions. “Many drugs provoke paradoxical responses,” he says. “These are effects that, once understood, could be applied in industry. It’s time to rethink some of these underlying concepts.” (For an initial outline of the new approach, see “A New Prescription for Drug Development.”)
Such an integrative approach will require more interdisciplinary collaboration. A key component of the therapeutics initiative is a new Laboratory of Systems Pharmacology (LSP), a therapeutics research facility. Additionally, a therapeutics foundry will aim to develop methods and technologies for smarter drug design—for example, using known molecular parts to build proteins with desired functions. The LSP (construction is scheduled to finish in the spring) will eventually house an estimated 60 researchers from Harvard, MIT, and Tufts, and hospitals including Massachusetts General, BWH, and the Dana-Farber Cancer Institute. Researchers will have similarly varied backgrounds, ranging from experimental to computational biology, and from basic to translational to clinical research, all working in a single physical space. “Colocalization drives interdisciplinary science,” says Sorger, praising the benefits of proximity among researchers. “No electronic technology we’ve discovered has been more than an aid.” Loscalzo, whose own lab employs both experimental and computational approaches to medically relevant issues, hopes that by working alongside each other, researchers will gain a deep appreciation for the power of different research methods.
The new lab will explicitly tackle complex problems like neurodegenerative or inflammatory diseases, where traditional drug-discovery methods have made little progress. “I don’t think these are intractable problems,” says Loscalzo. “We have the data sets. We have the cellular and animal models, and we know the biochemical, molecular, and cellular underpinnings pretty well.” Here, collaboration is crucial. He suggests that a better understanding of basic biology will enable clinicians to characterize disease profiles in terms of their underlying biology, rather than their large-scale, end-stage physiological effects. In many cases, Loscalzo says, “the therapies that have been used so far have been largely focused on the end result of a disease, not the causes.” Exploring those causes could lead to novel therapeutic targets, as well as more effective diagnosis and treatment in a clinical setting.
The initiative also aims to foster a more collaborative relationship between academia and industry. The high cost of drug failures places a limit on how much companies are willing to risk. “You get stuck in a rut,” says Sorger. “Research is too expensive, so you have to go with today’s ideas, even if today’s ideas aren’t good enough.” In contrast, he says, academia is better equipped to handle long-term, open-ended questions and to investigate principles that could lead to more rational drug design and usage. To that end, a graduate program in therapeutics will train students in the science behind drug discovery and regulation, while requiring internships at pharmaceutical companies to get a taste of industry. “Exposure to real-world problems will help students think about their own research projects,” he says, by showing them what topics are best suited to each context. Nine students from existing HMS medical and doctoral programs are expected to enroll this fall in the new therapeutics certificate program.
Sorger also sees a role for academia in mediating the adversarial relationship between pharmaceutical companies and federal regulatory agencies. Regulatory science, he says, could be restructured to enable companies to alter and improve their treatment regimes during the trial process, and to continue monitoring after a drug reaches the market. “The FDA is complicit in the reductionist view of drug development, in that the approval process requires the pharmaceutical industry to identify a specific target for the drug candidate,” adds Loscalzo. Yet infrequent toxicities and nontoxic side effects are also important components of how clinicians prescribe drugs. Furthermore, combination therapies may be the way forward for drugs like Vemurafenib (new drugs are already in development to combat the observed resistance), but the current lengthy approval process discourages collaboration between industry competitors on potentially powerful drug cocktails. Plans are under way for a partnership between the new therapeutics initiative and the FDA to add nuance to the current regulatory structure, and to implement a structure for failure analysis, as in engineering. Sorger also hopes to develop the science needed to test novel treatment methods; gene therapies, stem-cell therapies, and engineered proteins, for instance, are promising research frontiers that the current system is poorly designed to evaluate.
Ten-year costs for the therapeutics initiative are estimated at $200 million, with significant funding anticipated from private and philanthropic as well as federal sources; a $5-million grant from the Commonwealth’s Massachusetts Life Sciences Center is funding the construction of the LSP. In the next decade, Sorger believes, the initiative will make significant advances in areas like toxicology and personalized therapy. Although federal budget cuts have drastically decreased funding of scientific research, Sorger is undeterred. “In crisis lies opportunity,” he says, and institutions have been willing to consider more collaborative ways to organize research. The financial crisis and public debates on healthcare have imparted an additional sense of urgency to current research. “Given these tough economic times, people realize that we are fundamentally dependent on the success of the broader economy, and our economy is, in part, medical practice,” he says. “The piece we can drive is innovation: innovation focused on improving patient outcomes and reducing costs.” By promoting a more integrative view of drug development—from research through testing to regulation—the new therapeutic science team hopes to provide the needed change in the status quo. “We’re trapped in a linear narrative here,” says Loscalzo. “Genetics, genomics, and conventional wet-bench biology have evolved through linear, reductionist reasoning. It’s not a feasible approach if you’re thinking about systems and networks.”