A Strategy for Making Clinical Trials More Successful

The world's tallest structure, Dubai's Burj Khalifa, cost $1.5 billion to build, which is roughly half the average cost of delivering a new medicine to patients. The price tag for new drug approvals has risen to roughly $2.6 billion, and $2.9 billion if you count the expense of required post-approval studies. The cost of success is so enormous because of all the money invested in drugs that fail: 90 percent of test drugs that enter clinical trials never get approved.

"The single biggest category of spending in R&D is clinical development, and the overwhelming majority of money spent on clinical trials goes toward programs that fail," says Rob Lenz, senior vice president, Global Development. "As an industry, we need to increase the amount we spend on drugs that reach the market and decrease the amount we spend on those that don't."

To tackle this challenge, Amgen has placed a large bet on human genetics, which can help to pinpoint drug targets that are more likely to lead to new medicines. More recently, another promising strategy has emerged that could also boost success in the clinic by changing the way that clinical trials are designed.

"Many of us think of innovation as something that only happens in the laboratory," said Lenz. "Clinical research is seen as more of a box-checking exercise, where we run fairly cut-and-dried randomized, placebo-controlled studies based on established protocols. But things are actually changing very rapidly in the world of drug development. Truly innovative methods for clinical trial design are maturing or mature."

Moving from fixed to flexible study designs

In conventional clinical trials, the protocol of a study is set in stone before the first patient is enrolled. The criteria for the patients being enrolled, the doses tested, the outcomes measured, the study's duration—all these things are established up front, and the protocol is executed without change.

Rob Lenz, Senior Vice President, Global Development, Amgen

Newer methodologies, like adaptive clinical trials, are far more flexible. "With adaptive designs, you can monitor the incoming data and modify the protocol based on what you're learning as the study unfolds," Lenz noted. The potential changes need to be spelled out before the trial begins. If the pre-established criteria are met, you can make a range of adaptations, including

• Dropping or adding doses. Eliminating doses that aren't effective, differentiated, or safe can increase the value and efficiency of the study by allocating more patients to the more informative doses.
• Increasing the size or duration of a trial, which gives the test drug a better opportunity to demonstrate its true impact.
• Enriching the study population by adding more of the types of patients who respond to the treatment being investigated.

In addition to generating greater clarity around a test drug's potential, adaptive designs can also benefit patients. "The sooner we can be confident that a drug or a dose of a drug either works or doesn't work, the sooner we can either advance that drug or stop evaluating it in patients," said Lenz. "To me, that's a great example of patient centricity."

Amgen's R&D strategy highlights the role these approaches can play in accelerating drug development and increasing the probability of success. Amgen established the Center for Design and Analysis (CfDA) to evaluate advanced trial designs and perform the modeling and simulations needed to support this work. The center includes new departments focused on Data Science, Design and Innovation, and Biostatistical Science.

"We decided to centralize these capabilities so that they can be deployed to all product teams, not just the teams that already happened to have this capability," said Lenz. "These methods can give us the right amount of information to make a decision—not too much and not too little."

Amgen has doubled down on the new capabilities by launching a major culture and capabilities-focused initiative to accelerate their adoption. "We see these methodologies as critical to our mission of serving patients," said Brian Bradbury, a vice president who leads Amgen's Center for Observational Research. "By reducing spending on programs that will ultimately fail, we can pursue more promising programs than would be possible under the traditional model for clinical research."

Improving speed and success while reducing cost

In conventional clinical development programs, tradeoffs are required if you want to prioritize speed, or cost, or likelihood of success. To optimize for success, you often accrue more data at each stage of the program, but that makes the program go slower. Optimizing for cost by staging your investment also slows you down, while optimizing for speed tends to drive up the cost. With adaptive designs, it's possible to simultaneously increase the odds of success, reduce spend, and get to an answer more quickly.

Adaptive designs can boost the odds of success by correcting assumptions that may otherwise obscure a drug's true potential. "In designing any clinical study, we need to make assumptions," Lenz observed. "We assume the patients will look a certain way; the disease will progress at a certain rate; the doses will perform a certain way; the treatment effect will be a certain amount. Invariably, some assumptions are wrong. That's why we're running the trial—to learn these things.

The risk is the wrong assumption can sink a drug that might otherwise have succeeded. "It's hard to say what proportion of drug development failures are due to avoidable flaws in trial design, but it's probably substantial," said Lenz. A few years ago, the FDA published an analysis of the agency's reasons for delaying or denying approval to 150 new drug submissions received from 2000 to 2012. That analysis pointed to "several potentially preventable deficiencies, including failure to select optimal drug doses and suitable study end points" as major factors in failures and delays.

By providing a mid-course assessment of a clinical trial's performance, adaptive designs allow for course corrections that may lead to improved results. "In the past, a lot of failures were due to picking the wrong drug targets," Lenz noted. "As we get better at picking targets by using human genetics, clinical trial design will become the next variable we need to optimize to improve success rates."

Making adaptive designs Amgen's default approach

While the concept of adaptive clinical trial design has been around for some time now, adoption of the new paradigm in the industry has been slow, at least until recently. The new methods were initially seen as novel and risky compared to familiar methods. "At Amgen, we've come to recognize that there's a much larger risk in not deploying methods that can get medicines to patients faster and more reliably," said Lenz. "All of these techniques can increase a trial's probability of success, and we intend to use them every chance we get."

Toward that end, Amgen has also enhanced its capabilities in the field of clinical trial modeling and simulation. Simulations can model variables such as the rate of patient recruitment, the time it will take for the drug to have an impact, the size of the impact, etc. The goal is not to predict the study's results, but rather to show how different study designs will perform under different scenarios.

"Clinical trial modeling is really no different from what you do when you want to shoot a new rocket into space," said Lenz. "You simulate all sorts of weather conditions, fuel consumption, and other variables to design a rocket that performs optimally under a range of conditions. Simulating clinical trials can't make a drug work if it isn't going to work, but it can help you design an optimized trial with the highest likelihood of providing meaningful and actionable answers."

Results from simulations can be surprising and counterintuitive, Lenz said. "As an experienced drug developer, you imagine that you can look at different design options and quickly pick the one that makes the most sense. With simulations, you may find that the design option that looks the worst is actually the best and may give you an answer faster, at less cost, and with equal or higher odds for the trial's success."

Adaptive designs are now being used in a number of mid-stage programs in Amgen's pipeline, including rozibafusp alfa (formerly AMG 570), a novel antibody-peptide conjugate under development for lupus, and efavaleukin alfa (formerly AMG 592), an IL-2 mutein Fc fusion protein that is being investigated for the treatment of lupus. Efavaleukin alfa is participating in the FDA's Complex Innovative Trial Design (CID) Pilot Program.

"The vast majority of Amgen's clinical trials now use modeling and simulation to inform the study design, and a large percentage of our studies now incorporate some adaptive elements," said Lenz.

Better data, regulatory openness

Several external factors are making adaptive trials more practical and attractive. The data needed for accurate simulations comes from two sources, both of which are becoming more abundant and reliable. They include real-world data collected from electronic health records and other sources, and patient-level data acquired in previous clinical trials. Data from previous studies have only become available recently through industry collaborations, including the Placebo Standard of Care initiative sponsored by TransCelerate. These data can be used to estimate the true amount of patient-to-patient variability that is likely to be seen in a planned trial. The resulting insights can help to predict when the data collected in a study will be mature enough to guide adaptive decisions.

The 21st Century Cures Act, passed in 2016, requires the FDA to provide updated guidance on how companies can use adaptive trial designs in a way that satisfies the agency's evidence standards. "Senior people at the FDA are very open to the idea of what they call simple adaptive designs, and there's also a growing openness to more complex adaptive designs," said Lenz.

Lenz added that Amgen's scale should give it an edge in the race to reap the benefits that adaptive trial designs can provide. "In very big companies, it's difficult to rapidly engage the entire organization behind a significant change like the adoption of adaptive design methodologies. Smaller companies can move more quickly, but they may not have the resources and expertise to make this sort of change. Amgen is large enough to put considerable muscle behind innovative methodologies, and small enough to move quickly and nimbly. So I'm very optimistic about our prospects to become an industry leader here."