The drug development pathway is not really working as intended at this juncture. There have been ample recent examples of drugs failing miserably in the late stage pipeline. Many investigators have highlighted a need for better biomarkers to allow for more discrete targeting of patients who are most likely to respond to a given therapy in an effort to cut down on these "mass market" failures. On the otherhand, critics have maligned the use of surrogate markers to get drugs approved faster and thereby compromise patient safety. And, we have fewer regulators to do the oversight job across a much wider global geography. What's a regulator to do?
We need regulatory innovation. About 3 weeks ago, I saw this piece on "A New Way to Approve Drugs" in the In the Pipeline blog. In it, Derek Lowe cited a new article on this topic in BioCentury in their Back to School issue - entitled Regulatory Innovation. I finally read it and was impressed with the concepts promulgated.
The article advocates for the adoption of an adaptive regulatory system. This is really needed in order for regulators to remain current with new science in reviewing and approving new drugs. It states flatly that the existing system delays putting new medicines and diagnostics into the hands of patients and doctors who must ultimately decide whether a treatment is worth the risks. The current system is also plagued by steadily rising costs and steadily decreasing introductions of new products.
So what would this adaptive system look like? We can look at the Orphan Drug Act, Children's Oncology Group (COG) and the activism for drug development characteristic of AIDs crisis in the 90's for inspiration. We need to configure the system to learn. While the exact components of such a system need to be fleshed out, at least four transformational initiatives will need to be launched to support it.
- Industrial scale system to identify, characterize and qualify biomarkers for regulatory use
- Restructuring the linear clinical trial path into a parallel throughput system coupled to a post-market system
- Developing a conditional approval system that speeds product to market (for qualified use) and has pre-specified criteria for moving it to full approval
- New incentives to develop targeted therapies - for drug/diagnostic co-development and for tackling tough diseases that can't be assailed in the current IP timeframes
Biomarkers are the starting point for a system that learns and adapts. We need more and better biomarkers with stronger correlation between endpoints and clinical benefits. Biomarkers enable us to target patient subpopulations that are most likely to benefit from a treatment - and thereby give both the patient and the drug maker their best shot at positive outcomes. They also help to eliminate patients who will not benefit from receiving the drug and prevent them experiencing unnecessary side-effects.
Biomarkers (surrogate markers) could also be used to demonstrate safety or efficacy with smaller and/or shorter clinical trials. But as yet, there are only 32 FDA-qualified genomic biomarkers (mentioned by about 10% of drug labels) - and we still need a widely accepted research paradigm for statistical validation [of a surrogate endpoint]. Biomarkers won't do it alone.
We need parallel throughput trials to back adaptive regulation. The randomized trial model was a great boon to drug development but it has begun to stifle innovation. Large randomized trials are too expensive and often take too long - and address only a limited number of questions. According to the BioCentury article,
"the clinical research enterprise is like a medieval workshop that turns out expensive, exquisite, and incompatible one-of-a-kind artifacts"
An adaptive system would support new models and recognize that public health is sometimes served by acting with speed on approximate data. We need an infrastructure that can learn more quickly and use statistical analyses that capture individual benefits in a broad patient array. In an adaptive trial, biomarkers can be used to assess responses to different investigational agents and that information can be used to optimize patient assignments in future patient trials. Thus patients are helped by those who preceded them and their experience can help those who follow.
Adaptive Phase II screening protocols can overcome several problems:
- Testing novel agents in first-line settings for diseases that can be treated with existing therapies
- Testing combinations of two or more unapproved drugs through collaborative parallel throughput testing systems - multiple compounds with multiple sponsors
- Because adaptive trials can more rapidly benefit patients, this benefit can help overcome issues with patient recruitment for clinical trials - by better answering what is benefit for me
We need a conditional approval system too if we're going to rely more heavily on biomarkers, adaptive designs, and speed approvals. Conditional approvals would be best applied to meet any of the following four health priorities:
- Diseases where no treatment exists
- Where existing treatments need to be replaced
- Where patients are not adequately served by existing treatments
- Where a drug is disease modifying
In exchange for quicker approval, there would be check on promotion, mandatory informed consent, and enrollment in electronic registries -- along with prespecified criteria to transition to full approval. Approval status would change as more information became known. There would be continued assessement of sensitivity and selectivity of biomarkers. Conditonal approval would have a goal of bringing improved treatments to patients more quickly and through it patient segmentation will become the norm. Regulators would consider the risks of NOT providing access to a treatment as well as the potential for adverse effects. We all need to adapt to the mindset that with conditional approval the information gathered in the postmarket is likely to change perceptions about a drug - a drug could turn out to be better or worse than anticipated.
Finally, we need incentives. As an example, The Orphan Drug Act sparked an explosion of new drugs for rare diseases - literally going from a dozen or so to more than 350 since its enactment. A similar exclusivity model could be applied here to induce companies to pursue diagnostic-therapeutic combinations to target specific patient subpopulations. This type of approach could help us get more surrogate endpoints developed to speed conditional approvals. Similarly, incentive approaches could help get more companies interested in demonstrating long-term imporvement with conditions like Parkinson's Disease - where the time required can exhaust any patent term remaining. Extended market exclusivity might be such an incentive in a case like this.
Adaptive Regulation can bring much needed innovation to make game changing improvements to the current paradigm. This can get us out of "tinkering mode" with R&D and business models, or in concentrating on geographic expansion to emerging markets - that don't provide true innovation.
In the 90's, AIDs advocates pushed the medical community to innovate - with things like drug approvals based on surrogate markers and parallel track clinical trial designs. Lack of progress in addressing needs like diabetes and Alzheimer's constitutes a silent crisis. It's surprising with millions of Americans facing those two conditions, that the AIDs model was not more readily replicated with other diseases.
But there is no reason that it cannot be.
Posted by Bruce Lehr October 11th 2010.
