June 2011
The current blog will explore possible interpretations of some interesting new analyses reported through Nature Reviews Drug Discovery on why drugs fail in Ph. II. (Nature Reviews Drug Discovery, volume 10, May 2011, 1).
Setting aside for the moment that this is self-reported (blind) data from a select set of large pharmaceutical companies, there are some nuggets in the short paper worth exploring, given that it describes an extraordinary collapse of success rates over a short period of time.
The core of the discussion is based upon an “analysis by the Centre for Medicines Research (CMR) of projects from a group of 16 companies (representing approximately 60% of global R&D spending) in the CMR International Global R&D database [which] reveals that the Phase II success rates for new development projects have fallen from 28% (2006–2007) to 18% (2008–2009), although these success rates do vary between therapeutic areas and between small molecules and biologics. As the current likelihood of a drug successfully progressing through Phase III to launch is 50%.” (Nature Rev. Drug Discov. 10, 87; 2011)
Beyond the overall assessment, the following sub-analysis is also interesting: “… 51% (44 out of 87) [of Ph. II failures] were due to insufficient efficacy, 29% (25 out of 87) were due to strategic reasons and 19% (17 out of 87) were due to clinical or preclinical safety reasons. Out of the 25 failures that were terminated for strategic reasons, 16 involved validated targets … suggesting that some of these failures were due to inadequate differentiation from more advanced drugs in the same class or from drugs with similar indications in another mechanistic class …”
The paper’s authors rightly conclude that: “Although it is difficult to draw conclusions from these data, the finding that a substantial proportion of Phase II failures were due to strategic reasons suggests that one important underlying factor could be overlapping R&D activity between companies with drugs in Phase II trials. This raises the question of whether an increase in collaborative efforts between companies up to the point of proof-of-concept for novel targets or mechanisms might be more cost- and time-effective.”
To be sure.
However, they could easily have taken this a few steps further. There are three ways of interpreting the analysis:
Good news (from a strategic & financial perspective, Panglossian as it may be)
Bad news (from a technical and regulatory perspective)
No news (from an opacity of perspective, perspective)
First, the “good news”:
If 50% of failures are due to early efficacy signals (or a lack thereof), while only 20% are safety associated, it suggests that the industry is doing an extraordinary job in medicinal chemistry, preclinical toxicology and phase one. We apparently now have very clean drugs coming through (after all, classically speaking Phase II is supposed to be a safety review and one would expect a relatively high percentage of safety related failures). This probably also speaks to the high percentage of antibodies in the mix, and possibly the high volume of cancer trials in the mix (as you are more willing to accept higher toxicities in cancer). While it is difficult to compare this analysis with prior assessments, were one to go back 15 years, I’m sure that the percentage of failures due to unanticipated safety signals would be substantially higher.
By picking up early efficacy signals (which, by the way, implies that many Phase II programs have now moved to randomized control trials at this point, an aggressive evolution of the overall industry R&D structure which makes those programs considerably more expensive, but considerably more useful) you are avoiding the time and extraordinary cost of running a disastrous Phase III: Earlier, quicker kills. Driving to early quick kills has been the industry’s rationalization mantra for at least the last five years. If the above interpretation of the data is near the mark, the industry’s efforts have not gone unrewarded.
Perhaps even more importantly, consider the economic value of avoiding a me-too launch. As noted: that Phase II trials are relatively small. If one is seeking a meaningful efficacy signal in a very small trial then this implies that one is expecting to see a very large improvement v. control (standard of care, placebo or baseline as the case may be). If a firm is willing to cancel a program that provides no obvious and substantial benefit to patients early on, the firm avoids the risk of running a very large Phase III trial that results in a statistically significant but clinically meaningless (and financially ruinous) outcome. The data can be, in one light anyway, interpreted as a demonstration of strategic discipline on the part of the pharmaceutical industry.
Now the “bad news”:
The extraordinarily low and dropping Phase II success rates overall may suggest that as the comparator standard of care improves, on average it gets sequentially harder to make improvements. A red queen effect has begun to tighten its grip on the industry (a commonly used co-evolutionary metaphor from Lewis Carroll’s Through the Looking Glass – the Queen forces Alice to run faster and faster just to remain in place). At the moment, an increasing amount of ingenuity, insight, information and investment is required to take ever smaller strides forward. This effect is dramatically represented in any historical plot of declining industry success rate : dollars invested.
To use another metaphor: If you consider the current investment environment as a fitness landscape, it becomes clear that something very disruptive, either scientific (a dramatic leap forward in our understanding of biological processes), technical (radically new platforms), or mathematical (a vast improvement in the efficiency of trial design) is required to move to a higher peak.
Further, if the state space illustration below is a reasonable representation of potential, finding a higher peak is in no way a given (a move to future state B (lower fitness) is as likely is a move to future state A (higher fitness)).
Finally the “no news”:
It is possible to assume a somewhat cynical perspective on this entire analysis. Consider the following, if the MTD of my drug is insufficient to drive a therapeutic response, is that an efficacy problem or a toxicology problem? It’s a bit of a glass half full vs. half empty story; or more specifically… a therapeutic window half open vs. half closed story.
It certainly sounds better if I report that I can’t get efficacy with my chosen dose, vs. I can’t get the dose to a therapeutically useful level without killing the patient (remember this is Phase II after all, where I’m doing dose ranging studies). This, by the way, doesn’t require any conspiracy on the part of the industry to misinform. Research will develop an MTD hypothesis in preclinical and Phase I studies. Once one or perhaps several potential doses have been selected the drug is tested through Phase II. When the data suggests a lack of benefit, the result is reported as lack of efficacy. But the history of the drug has been lost at that point. The dose selection may well have been decided upon because of early toxicology concerns. So is it a toxic drug or an ineffective one? The answer in many cases is, probably a bit of both. This nuance is lost completely in the Nature article and CMR analysis.
Finally, a word on discontinuation due to strategic re-prioritization (29% of failures in the dataset). That’s a terribly ambiguous category to say the least, although that said, our own analysis (Clinicaltrials.gov) suggests that only about half of programs that publically declare Phase II success actually go on to initiated Phase III trials. Considering what might lie below the surface (and this is only a partial list):
Poor decision analytics and execution
Poor Target Product Profile development
Long delays in development
Disagreements with the FDA as to how to proceed forward
Bad luck
Rationally or irrationally betting on the failure of others at a later or similar stage
Unanticipated pivotal requirements (e.g., changes in stance of FDA)
Unanticipated REMS
Undisclosed safety signals
Design
There was never any intention to carry the program forward, but instead the trials were run to prove the mechanism in a small population (e.g., assessing the utility of disrupting a signaling pathway in a very rare tumor type only to move forward later an a distinct disease that is also thought to operate via that same pathway)
A declared “success” is primarily broadcast to generate partner interest, but which is then followed by discussions which unfortunately lead to an external invalidation of the result
While the above is conjecture, it is reasonable to state that few if any companies are going to “strategically discontinue” a program that has any discernable clinical value. Anything monetizable will be monetized. While the market may not be perfectly efficient, the more dire the economic situation, the more likely companies are to search beneath the couch cushions for a few coins.
Parting Thoughts
A numerically based assessment of that which has or has not come to pass is of limited utility, other than to suggest that things are perhaps not going as we might like. One can become addicted to the analyses, and the past is a very poor model of the future. More importantly, none of this answers the more pressing and fundamental questions:
Do the technologies exist today that will result in fundamentally more valuable target product profiles considering an ever advancing standard of care?
Do we have the enabling technologies that will allow us to bring trial sizes to economically manageable levels?
As we learn more about the pharmacological surgery that we are doing to ourselves, and are better able to undertake outcomes analysis, is the FDA likely to get less, or even more stringent in its requirements (e.g., diabetes, obesity etc.)?
The path forward is going to need to be disruptive. That means high risk, high capital intensity early on. That means government investment in pre-competitive work and the backing of market exposed economic vehicles. As the market-based argument for drug development evaporates, but the public health need remains, so this should be a call to action in Washington, and in the academic and quasi- academic research networks around the country and the globe. But…this is all for another blog entry…
Tweet
About the author: Nate Dowden is a Managing Director at the Frankel Group. To contact him regarding this post, email blog@frankelgroup.com.