December 1, 2007 (Vol. 27, No. 21)
Experience Matters When Designing and Executing Human Studies
Some 30 years after the birth of biotech, drugs are failing at higher rates than ever. For example, between the late 1990s and mid-2000s only eight percent of all oncology drug candidates in the U.S. won marketing approval, down from 20% 10 years ago. Meanwhile, drug development costs have escalated to an average of $1.2 billion, according to data from Tufts Center for the Study of Drug Development.
Biotech companies say that’s because the easy drugs have been developed. “Failure in science is absolutely fundamental to scientific advances. You’re wrong about 90 percent of the time,” insists Larry Meinert, M.D., senior vp for medical and scientific affairs at Covance (www.covance.com). And, he adds, the more innovative the therapy, the more likely it will fail.
That’s true, but it’s not the whole story. There’s a difference between scientific and operational failure, he points out, and there are several strategies biotechs can use to improve their success rates.
One of the major reasons clinical trials fail is that trials are often initiated before the fundamentals are in place, according to Allen Cato, M.D., Ph.D., founder of Cato Research (www.cato.com).
Brian Huber, Ph.D., vp strategic drug discovery at Quintiles Consulting (www.quintiles.com), says that one of the fundamentals is to begin discovery and development with a true picture of what you’re trying to develop. “You have to ask many questions,” he says. “What is the real unmet medical need? What do patients need? What would a physician prescribe? What would the FDA approve? What would third-party payers reimburse?”
Quoting Nobel Laureate Trudy Elion, he says, “You don’t try to sell what you discover and develop. You discover and develop what you can sell.”
“Ten years ago,” Dr. Huber continues, “we looked for novel compounds. That’s not enough. Now, we look for novel medical outcomes compared to the standard of care, and the drugs also must make pharmacoeconomic sense. Even successful companies are still trying to understand that.”
Companies, however, are still rushing to the clinic. According to Yan Lavrovsky, M.D., Ph.D., executive director at ChemDiv (www.chemdiv.com), the value of companies is based more on the value of their drug development pipelines than on the outcomes of Phase II or Phase III trials. Consequently, he says, “bad molecules are getting to trials because of pressure from investors.”
Regulators, ideally, should be partners in the trial design. “We often work with regulatory agencies before going into the clinic,” says Lynda Sutton, Cato’s COO and chief regulatory officer. “A lot of companies don’t do that.” Big companies sometimes rely on their internal resources and experience and don’t engage the FDA early in the process, Dr. Cato adds, “so they don’t ask the fundamental question, What does the agency want?” and go forward with trials. “They’re taking a very big risk,” Sutton says.
“Given the huge cost of drug development, you have to make every buck you spend have impact,” John Ho, M.D., senior vp, corporate strategy, Charles River Laboratories (www.criver.com), emphasizes. Talking with regulators before the trial is designed helps avoid conducting studies that aren’t valued by the regulatory authorities. “In some areas, the FDA endpoints aren’t as clear-cut as one would like,” Sutton says.
During a clinical trial about four years ago, Dr. Cato recalls, oversight responsibilities shifted when a new FDA division was created. The regulatory team now reviewing the trial data envisioned different endpoints than the original FDA team and wanted an additional, much larger Phase III trial. Dr. Cato, however, had coordinated a Special Protocol Assessment with the original team, and had it in writing. Because of that, the originally agreed endpoints were accepted, and the FDA deemed a second Phase III trial unnecessary.
Such coordination can also inform the choice of models. This is an iterative process, going from preclinical models through clinical studies and then back to the models to assess their predictability and their future utility, Dr. Huber suggests. “Animal and human models have a good correlation for some diseases like blood pressure,” Dr. Lavrovsky adds, “but others, like depression, are less straightforward and are difficult to model in animals.” Therefore, research in such areas constitutes proof of concept and safety’ studies and then moves quickly to human trials.
To learn which models are likely to be predictive for a given compound, Dr. Huber recommends testing a class of drugs on a variety of models to develop confidence in specific models and build a knowledge base that can be referenced over time. “Draw conclusions about the model, as well as about the compound,” he advises.
“Companies are seizing on sequential design as a solution to clinical failures,” Dr. Huber continues. “That’s a great way to narrow the field of possibilities, but results in making fundamental decisions based on less and less data.”
Additionally, Dr. Meinert says, despite ever-increasing knowledge of receptors as specific targets of interest, mechanisms of action are less well-known and researchers tend to approach receptors one at a time. Instead, he says, “we need to hit a series of receptors.”
“Statistical problems should never make or break a drug,” Dr. Huber says, yet they do remain an issue. The problem, he notes, is that researchers prospectively design a study based on their assumptions regarding efficacy, safety, and the like, and the trial is also designed based upon those assumptions. Then, if the actual data doesn’t match some of those assumptions, the expectations weren’t met.
It’s important, Dr. Lavrovsky notes, to design trials that closely mimic real-world situations. Patients, he says, unlike lab animals, aren’t pretreated, and often have undergone all the traditional therapies before enrolling in trials. Therefore, CROs often recommend a polypharmaceutical approach throughout trials, “but preclinical work that’s focused on shortening the time to begin human testing doesn’t include that,” Dr. Meinert says, “even though more preclinical testing could shed light on what may happen in late-phase trials.”
Traditionally, in the early phases, patients in trials have advanced stages of the disease and have undergone several conventional therapies. “That’s not the best setting,” Dr. Huber says, suggesting that companies conduct less single agent, salvage-type therapy and treat patients earlier in the disease process, and add their drug to a traditional therapeutic regimen.
Know your drug, the CROs overwhelmingly say, to further chances of success. Oftentimes, Dr. Cato elaborates, people look at endpoints, expecting to see effects within about two weeks. “For some compounds, though, they have to treat for three weeks” to determine whether the drug is working. Consequently, good compounds are discarded because the trials aren’t long enough.
Similar situations arise with doses. Nowadays, companies treat to the maximum tolerated dose, but “dose-response curves often are bell-shaped,” Dr. Cato says. Consequently, doses are only effective at the curve’s mid-levels.
“Companies, quite often, don’t do precise patient certification,” Dr. Lavrovsky adds. For some clinical trials, this can be very important, affecting outcomes. But the issue goes beyond that.
“Patient accrual can be a problem in some areas,” Dr. Cato reports. Several companies are competing for the same patient population and the number of investigations has declined. “Some 83 percent of studies are against common diseases,” Dr. Meinert says. “Thanks to the biotech sector, everybody in the world who cares to, understands which targets are next.” Simultaneously, the capital costs of synthesizing compounds have dropped dramatically “at any given time 20 to 50 companies are pinging these receptors and patient groups, versus 5 to 6 companies several years ago.”
For investigators, participating in clinical trials has become uneconomical “and not worth the hassle,” Dr. Meinert states. “Two-thirds have backed out, in favor of saner hours, an improved work/life balance, and a lot less paperwork.”
Phase III Failures
Phase II is the most critical part of clinical development, and “Phase III should be the most boring if your Phase II trials are done correctly,” Dr. Huber says. “If done correctly, there’s a 70 to 80 percent chance Phase III will be successful. Problems arise when the Phase II studies do not accurately identify the proper dose, schedule and patient stratification for the Phase III studies.”
Many late-stage failures occur, Dr. Huber says, because “companies don’t conduct innovative Phase II programs that critically test the ability of their compounds to meet the minimal target product profile.”
Instead, Dr. Meinert adds, companies sometimes too narrowly define patient populations in Phases I and II because it’s easier and then get surprises in Phase III that affect efficacy. The caveat, he says, lies in the population variability inherent in studies of thousands, versus hundreds of patients, which increases the chances of identifying something not seen in smaller trials. “You can never avoid that,” Dr. Huber says.
Additionally, at this stage, drugs often are pulled not because of scientific failure, but because they have no significant commercial advantage in terms of efficacy, reduced side effects, or dosing. “A lot of drugs fail in Phase III because the wrong dose was chosen. The choice of 5 mg versus 7.5 mg can make the difference between a drug making it to the marketplace or not,” Dr. Huber says.
“If a drug fails in clinical development, it’s important to know whether the drug failed or whether the target failed,” Dr. Huber says. “In other words, did the drug not reach the target at the proper concentration for the proper time or did interfering with the target not produce the desired medical effect? Design a trial that suggests which failed.” For example, for a central nervous system drug, he advocates using PET scans in the trials. It increases costs and time for the study, “but you can make informed decisions if the drug doesn’t work, and thus add significant value.”
Role of CROs
“Small biotech, in particular,” Dr. Ho says, “should consider working with an integrated provider who acts as a partner. Biotech firms may have great ideas and technology, but they often lack experience and infrastructure to go through preclinical development and early clinical trials effectively and efficiently.”
In that area, CROs have proven abilities. Because CROs generally have conducted more development programs than most biotech or pharma companies, they can offer best practices solutions and insights for IND-enabling packages and clinical trial design and execution.
“The area of receptors is one example,” Dr. Meinert comments. CROs have the greatest exposure to a wide range of receptors, with data in both animals and humans. Although the details may be proprietary, they can use that knowledge “to give guidance, although not specific information” to steer research away from probable failures and toward more likely successes.
“Don’t try to micromanage the CRO,” Dr. Meinert advises. Instead, he recommends a division of focus that lets the CROs handle the study’s logistical concerns and allows the sponsor to concentrate on the longer term implications of the results.
Some CROs have formalized their own internal separation of focus by developing consulting arms to augment their research expertise. Quintiles takes this a step further by including consulting as well as an investment arm called NovaQuest that, “in certain cases, may even help fund some of the studies in exchange for some of the rewards,” Dr. Huber says.