Drugbaron Blog

May 20, 2015 comments

The perils of ignoring negative data

This week, Eleven Biotherapeutics ($EBIO) announced the failure of its Phase 3 trial of an IL1 antagonist in dry eye.   The company indicated they saw no way forward for the product candidate in this indication, which with a little translation, indicates a comprehensive failure of the study (rather than a near-miss on the primary end-point, for example). Unsurprisingly, the company lost most (around 75% in the first few days after the announcement) of its value as a result.
Drug discovery and development is hard, so failure is inevitable. It’s easy, then to shrug your shoulders, declare this as “one of those things” and hope the rest of your portfolio fairs better in the grand pharmaceutical lottery.
But is investing in pre-approval therapeutics company just a lottery?   After all, when these companies come to market in an IPO (in the case of $EBIO in February 2014), there is usually quote a lot information already known. How predictable was this failure.
After all, in 2013 Eleven Biotherapeutics completed a phase 1/2a trial of EBI-005 in dry eye patients – a trial that, according to the company “informed the design of the phase 3 study”. This study showed the agent to be safe and well-tolerated, but while it showed a numerical decrease in various markers of dry eye, as well as an improvement in the patients view of the symptoms, these changes were not statistically significant.
In other words, the only prior study of this drug candidate in this disease showed no changes at all that could not just as easily have been attributed to chance.
Even the rationale for adopting this mechanism of action in dry eye was “possible” or “plausible” rather than compelling. Yes, IL1 levels are increased in tear fluid from patients with dry eye, but so are lots of other pro-inflammatory markers (such as TNF-alpha). There was little in the way of genetic evidence to link this pathway to dry eye in humans, and non-human models of inflammatory diseases are notoriously unreliable.
On the positive side, the market for an effective treatment for dry eye may prove to be substantial, as the disease is both prevalent and poorly served by existing treatment paradigms. But optimism about commercial potential is of no value unless the drug actually works.
Its easy to sound smart after the event, but investors in companies like $EBIO have to ask themselves if they are surprised at all by today’s developments. If they are, perhaps they need a primer on statistics (and in fact, the situation is worse than they might imagine, because the chance of success in the phase 3 study depends not on whether the earlier trials were positive but on the false discovery rate). Even compounds with positive Phase 2 data are less likely to also be positive in phase 3 than most investors imagine.


But faced with a plausible but unvalidated mechanism of action and negative study data, why would you invest?


None of this would matter very much if the Eleven Biotherapeutics tale were an isolated event. But its not. Prosensa ($RNA) and Celladon($CLDN) are two other companies from the recent class of biotech IPOs, who progressed candidates lacking in robust early-stage data into later stage trials funded by the recent public-market enthusiasm for healthcare – and failed.
Arguably, they are not just unlucky draws in the pharmaceutical lottery, but predictable outcomes.
This habit of conveniently ignoring negative data is not restricted to biotechs approaching the public market. It is endemic in our industry. Larger companies, with many products in development do it all the time: look at the anti-amyloid antibodies tested in Alzheimer’s Disease, such as bapineuzumab and solanezumab, which failed to show benefit in phase 2 but were progressed into huge, expensive phase 3 studies (all of which, to date, have failed) all the same
Perhaps public market investors looking at new IPOs have become so used to seeing negative data being brushed off, usually behind the convenient smoke-screen of post-hoc data analysis (such as finding a positive “sub-group of responders” among the sea of negative data) that they no longer see previous failure as any kind of a red flag?
That, of course, begs the question as to why its become so trendy to ignore negative data. Some of the reasons are obvious: if investors are willing to be hoodwinked so easily, then why not carry on? After all, you never know something positive may turn up by chance alone.  Along similar lines, many of those peddling failed product candidates understand the perils of post hoc data analysis about as well as the generalist investors they are talking to. They genuinely believe that they still have a good chance of success. The blind are leading the blind.
Other factors probably dominate outside the setting of an initial public offering. Pharma companies continue to pursue anti-amyloid antibodies because of the power of the long-held mechanistic hypothesis (a phenomenon DrugBaron termed “idea bubbles”). Something similar probably underpinned the phase 3 failures with CETP inhibitors. Because biology is complicated, if you are sure you are right then there is always a way to explain negative data – any way to explain it, other than recognizing the flaw in the underlying hypothesis.
The power of the in-built cultural bias towards continuing rather than killing projects also no doubt plays its part. In large companies in particular there is often a huge systemic push towards finding a way to continue (no matter what data has been collected) rather than to crystalise the loss on your watch. Everyone recognizes the danger of the sunk-cost fallacy when they see it in the behavior of others, but seem much less able to see it in their own behavior.
All his matters because the accumulating, and most importantly avoidable, failures in the new generation of public biotech companies undermines the pitch of the next cadre. Eventually generalist investors will spot the trend – and sooner rather than later as the examples, and the losses, pile up. And when that happens the IPO window the sector has enjoyed for two years will slam shut. Because generalist investors, it seems, struggle to differentiate a reasonable proposition (which, of course, may still fail) from a blind gamble based on negative data. Having been reminded of that often enough, they will cease to support even the deserving cases.
When that happens we will all suffer the consequences of the actions of those (well-meaning or otherwise) who chose to ignore the writing on the wall.

Yearly Archive