Please login to the form below

Not currently logged in

Never-ending story: modernising clinical trial design

The recently launched I-SPY 2 trial in the US is a showcase for adaptive trial design in drug development, with the explicit aim of improving the poor success rates for new cancer medicines in phase III testing
Spiral stairs

Recognition that the pharmaceutical development process needs to be modernised to tackle diminished R&D productivity is driving wider adoption of adaptive clinical trials, featuring flexible protocols in which preliminary results are used to modify the trial's course.

Adaptive studies introduce a flexible protocol which can, for example, accommodate adjusted parameters such as dose selection, dropping or adding arms and modifying accrual rates, all based on criteria laid down in advance. In general they address more than one question, not simply whether an experimental therapy is better than control as is the case with traditional trial designs.

The overall aims are to reduce trial duration and cost and reduce the high attrition rates seen in phase III trials, which are the most expensive part of the drug development process. Overall, phase III failure rates are 45 per cent or more across therapeutic categories, and rise to around 66 per cent for new cancer drugs.

"Adoption of adaptive designs in clinical trials has been particularly slow in oncology, in part because clinical endpoints in cancer tend to be long-term, and this means that few patients will experience an endpoint in the early stages of a trial and so be able to help direct its course," according to Don Berry, a statistician who specialises in adaptive trial design based at MD Anderson Cancer Center in the US.

Berry is one of the principal investigators in a new trial which arguably takes the adaptive trial design concept to a new height. I-SPY 2 employs a protocol that combines a number of adaptive features with the ability to test dozens of experimental compounds serially in patients with neoadjuvant breast cancer, including liberal use of biomarkers and imaging technologies to help identify and understand patient responses.

Control therapy in I-SPY 2 is four cycles of a taxane followed by four cycles of doxorubicin/cyclophosphamide followed by surgery. Experimental agents are added to the taxane portion of the standard therapy.

Unusual design
I-SPY 2, which is also led by University of California, San Francisco (UCSF) breast cancer specialist Laura Esserman, is unusual in that it has been designed wholly on the basis of predicting how well a compound will perform in phase III testing. For example, it can provide critical information firstly on whether a drug should graduate to phase III at all, and for those that do inform investigators in appropriate dose-selection and patient subset, based on biomarker signatures.

"One of the main reasons for pivotal trial failures in cancer to date is that they have enrolled the wrong patient population," notes Berry.

With that in mind, one of the cornerstones of the study is adaptive randomisation; a concept which has been pioneered at MD Anderson. In essence, this means that women put forward for inclusion in the study will undergo a biomarker screen to identify the most appropriate treatment arms.

Once the trial is underway, that decision is weighted to take into account the experience of prior patients with similar signatures. The trial subjects are still randomised, but with an increased probability that they will receive a therapy with higher response rates for their signature, and a decreased probability of receiving treatment which has performed less effectively.

Multiple treatment arms
Speaking at the launch of I-SPY 2 in New York, Esserman said: "We have set up a system where everyone can learn faster and, together, we can dramatically reduce the amount of time and the cost to bring those drugs to market that can make a difference in whether women live or die."

Of course, the design also means that it is possible to stop treatments quickly for futility, specifically if there is no subset of the disease which has at least a 10 per cent chance of success in phase III, according to Berry.

Another key element of I-SPY 2's design has been the inclusion of multiple treatment arms. In a kind of never-ending story, the study was set up with two drug treatment arms and a control, but others have already been added and there is no limit for the number of drug candidates which could be included.

"Provided patient accrual is reasonable, and at the moment it is going very well, we will be able to replace drugs that are dropped or which graduate to phase III," he said. All told around 20 compounds have been proposed by potential sponsors, although some may not make the cut. Additional biomarkers may also be included at a later date to expand the trial's scope and broaden the classes of compounds which can be included.

n = fewer patients
The adaptive design also has a dramatic effect in reducing the number of patients who need to join the study, perhaps reducing the total to 300 versus 10 times that many for a traditional trial. Futility decisions can be made after as few as 20 have been accrued, while efficacy signals require a few more – around 60 at minimum – to be recorded.

On average, if the drug has no effect on a disease subset this can be picked up with around 70 patients, but this is still pretty low for a phase II trial, according to Berry. And while the relative complexity of the adaptive design can add cost in phase II, for those drugs that graduate to phase III significant cost-savings should accrue from better selection of patient entry criteria. In effect, the pivotal trial is not diluted with non-responders, potentially rendering studies an order of magnitude lower in terms of patient numbers.

Greater industry acceptance
The benefits of adaptive trials seem now to be accepted by the pharmaceutical industry as a whole. Berry had been trying to get such a study underway in metastatic breast cancer for around a decade, but ran into early resistance from pharmaceutical companies that were reluctant to include their compounds alongside those of competitors in the same study.

"Times are changing, and at the highest levels in pharmaceutical companies there is a recognition that cancer drugs aren't going to work for all patients. That means they have to understand who will benefit, and which drugs will work in combination.

"Comparing them in the same sandbox is the only way to go," according to Berry, who notes that this is particularly true in cancer but also applies in other indications such as pain management, HIV and Alzheimer's disease.

I-SPY 2 has already brought together compounds from Pfizer (neratinib), Abbott Laboratories (ABT-888) and Takeda Pharmaceutical Industries (AMT-386), and there appears to be an almost unprecedented level of enthusiasm among the companies for the collaborative concept. In fact, other drug developers whose compounds failed to be selected have provided sponsorship and support for the study regardless, probably because they want to gauge the effectiveness of the approach and – perhaps – look at how other compounds in the same class fare.

It is also notable that industry has only contributed a small proportion of the overall costs of running I-SPY 2, with the bulk of the money coming from philanthropic donations. That in and of itself is an interesting departure from the traditional model of trials involving compounds originating in industry, and could have implications outside the confines of experimental therapy development.

Comparative effectiveness research
"One area of application that has gotten too little interest is what is called comparative effectiveness research (CER)," says Berry. "We are quite ignorant about who benefits from most therapies that are marketed and widely used."

Even though millions may use a drug, only some patients benefit while others may be over-treated and the I-SPY 2 concept could be adapted to apply to CER just as well as to the early development setting, according to Berry.

"A difference is that in the CER setting, funding is a major issue because there is little motivation for a company to run a trial whose goal is to determine which patients should not take their drug.

"In President Obama's last stimulus package he set aside $1.1bn for CER. This could help and it could yield economic benefits many times the cost. But I haven't yet seen an I-SPY 2 concept that is part of a CER project," he adds.

There is no doubt that momentum is building behind the concept of adaptive trials in drug development. In part this is thanks to the publication of the US Food and Drug Administration's draft guidelines on adaptive trials in 2010. These laid to rest concerns about how the agency would respond to marketing approval dossiers that included adaptive trial data.

Berry hopes I-SPY 2 will serve as a showcase for the power of adaptive trial design, and continue to encourage the pharmaceutical industry to adopt the approach. Aside from some beacon companies such as Eli Lilly and Novartis, which have substantial initiatives in the adaptive area, uptake has been somewhat patchy elsewhere.

"Adaptive designs promise great efficiencies in terms of resources and delivering better treatment to patients," says Berry.

"It's a real departure from the 'close your eyes, open them at the end' approach of double-blind studies, often followed by the statement: 'Oh rats, I wish I'd have looked where I was going!'"

Phil TaylorThe Author
Phil Taylor
is a freelance journalist specialising in the pharmaceutical industry.

To comment on this article, email




20th June 2011


Subscribe to our email news alerts

Featured jobs


Add my company
Streaming Well

Streaming Well is a healthcare-focused, award-winning video production company which operates in the US and Europe. We create engaging visual...

Latest intelligence

Influenza – the risk to vulnerable populations
Why we can't get complacent about flu vaccination...
Alzheimer’s Research UK highlights socio-economic inequalities in dementia risk
The charity aims to improve the number of women participating in dementia research and grow awareness of dementia risk factors...
Data security
Concerns about data security are building a strong case for clinical mobility in EMEA
Electronic medical records have transformed the storage of sensitive information but how can the healthcare sector continue to protect patient and staff data?...