Shares of Sarepta (NASDAQ:SRPT) are up more than 60% year-to-date based mostly on the progress of the company's exon-skipping Duchenne muscular dystrophy, or DMD, drug eteplirsen. On the surface, the drug appears to be a blessing for DMD patients.
Specifically, Sarepta recently announced that patients participating in the Phase IIb open-label extension portion of the study continued to show stabilization of their walking ability, assessed via a 6-minute walk test (6mwt). In the initial Phase II study, eteplirsen met its primary endpoint of increased novel dystrophin production at week 48. Based on these promising Phase II results, Sarepta announced that they plan on filing a New Drug Application, or NDA, with the FDA in early 2014. If approved, eteplirsen would be the first drug commercially available for DMD treatment, and is projected to reach blockbuster status within 2 years post-approval.
Investors' bullish outlook on eteplirsen went into hyper-drive last month when GlaxoSmithKline (NYSE:GSK) and Prosensa (NASDAQ:RNA) announced that their DMD candidate drisapersen failed to meet its primary endpoint in a pivotal Phase III trial. Despite showing promising Phase II results and increased dystrophin expression as well, drisapersen wasn't even close to outperforming placebo in terms of improving walking performance in a larger sample of DMD patients. This news sent Sarepta shares into orbit (up 25% in one day) because it essentially gives the entire DMD market to eteplirsen, pending approval.
So everything is good, right? Eteplirsen will be approved, etc…
While Sarepta is publicly spinning a compelling tale for investing in eteplirsen's prospects, the science tells a wholly different story. In fact, I am confident eteplirsen is not going to receive early approval by the FDA, and will in fact go on to fail in a Phase III trial.
The worst part is that the company seems to share my bearish sentiment, which is why they are pushing for early approval based on an extremely small Phase II trial.
To understand what's really going on, we need to look past the company propaganda. By contrast, the truth can be found in the dense, off-putting, boring scientific literature. Indeed, this is probably why the market has failed to recognize eteplirsen's imminent doom, and instead has pushed the stock to new heights.
With that in mind, I'll do my best to distill the important, and wholly misunderstood, aspects of eteplirsen. And subsequently, why the drug is actually in deep trouble and Sarepta knows it.
Disease, Therapies, and Treatment Outlook
Duchenne muscular dystrophy is an X-linked muscle-wasting disorder that presents in early childhood. The hallmark muscle weakness that characterizes DMD is caused by mutations in the gene that encodes for the protein dystrophin. Dystrophin is important to muscle function because it acts to stabilize muscle fibers during contraction by binding actin filaments to the extracellular matrix. Patients suffering from DMD have mutations in the large dystrophin gene that result in either a truncated, nonfunctional dystrophin, or a complete lack of dystrophin altogether, depending on the type of mutation present.
Boys afflicted with DMD are typically wheelchair bound by 12 years old, and usually die by the age of 20. There are no treatments for DMD. Instead, symptoms are managed using corticosteroids and intermittent positive pressure ventilation to help improve motor function, quality of life, and life expectancy.
Despite the promise of a host of gene therapies over the past decade, effective treatments for DMD have remained elusive because the mutation(s) underlying the disease are not universal across patients. In other words, DMD is a polymorphic disease requiring molecular therapies that are tailored to an individual's specific mutation.
Drugmakers looking to get the most bang-for-their-buck have initially focused on developing therapies for the most common mutations that are amenable to skipping exon 51. Even so, patients amenable to skipping exon 51 therapies still only comprise 13% of the entire DMD population, leaving ample room for multiple therapies. Duchenne's muscular dystrophy is in many ways a poster child for personalized medicine.
Sarepta's Exon 51 Skipping Drug Eteplirsen
Eteplirsen is a morpholino antisense oligonucleotide that skips exon 51 during pre-mRNA splicing of the dystrophin RNA transcript, resulting in a shorter yet functional form of dystrophin. The morpholino chemistry is believed to be therapeutically superior to phosphorothioate (drisapersen) because it allows significantly higher dosing without toxicity. Indeed, eteplirsen has been dosed up to eight times higher than drisapersen in human trials, which is thought to be the main reason for the increased dystrophin production.
Didn't eteplirsen significantly improve walking performance in clinical trials?
The simple answer is no. Despite multiple reports that eteplirsen did in fact show a clinically significant improvement in the 6mwt in a Phase II trial, the opposite is true.
Why did investors believe eteplirsen improved walking performance?
Again, the answer is simple. People didn't understand the statistics, and the company didn't exactly promote results that ran counter to their goal of selling stock.
The problem arises largely because the Phase II trial conducted by Sarepta only included 12 patients. Such a small sample size is problematic for a number of reasons. Chief among these is the fact that small sample sizes lead to low statistical power by default, and tend to have weird, non-normal distributions. And Sarepta's study was no different.
In their analyses including all 12 patients, eteplirsen failed to show a significant improvement in the 6mwt when the data was properly analyzed. Going directly to the actual peer-reviewed paper, it reports for the 6mwt at 24 weeks:
"As data distribution analysis revealed severe violation of normality assumption, the analysis was repeated using ANCOVA for ranked data, showing no significant differences between the 2 eteplirsen and placebo cohorts."
At 48 weeks, the paper states the following:
"As data distribution analysis revealed severe violation of normality assumption, the analysis was repeated using ANCOVA for ranked data, showing no significant differences between the 30 mg/kg and placebo cohorts but a statistically significant difference between the 50 mg/kg and placebo/delayed cohorts (p < 0.016)."
So it's a dosing and time issue, right? Eteplirsen works at the higher dose, correct?
While that would be a nice result, it's not quite right. The truth is that two boys in the lower dose group (30mg/kg) severely skewed the group's distribution. Sarepta researchers thus argued that the two boys should be excluded from further analyses because they are "outliers" due to their age, height, and rapid disease progression.
Regarding the minimally significant result in the higher dose, you should keep in mind that this result is not corrected for Type I error, which is explicitly stated in the paper. In other words, this "significant" result for the higher dose could very well be what's known as a false positive, and is probably not significant after applying Type I correction. No way to know with such a small sample size. But again, Sarepta hasn't gone out of their way to inform the general investor of such statistical issues...
What happens when the boys are excluded?
At 48 weeks, the 6mwt suddenly becomes highly significant (p < 0.0001) for the low dosage cohort compared to placebo.
So are the two boys really outliers? Which analysis is correct?
First off, I need to explain what an outlier means in statistical terms. An outlier is generally a datum that lies well outside the distribution of the overall dataset. Legitimate reasons to deem a datum an outlier include, incorrect data entry, suspicion of incorrect measurement, methodological problems, and so on. There are ways to specifically test for outliers and statisticians do tend to get grumpy when you don't have a good reason to include a particular datum in your analysis.
Why? Because excluding data that doesn't fit your hypothesis is tantamount to cherry picking. And unfortunately, that's what Sarepta did with their Phase II data.
Without explicitly performing an outlier analysis, they decided to exclude the two boys from the 6mwt and simply state that the boys were excluded because they skewed the distribution. Statistically speaking, that is non-sense. There must be a robust reason to exclude data, and normality of distribution isn't one of them. In the Discussion, the paper's authors go on to further justify excluding the two tallest boys based on their rapid disease progression, but again, this isn't a justifiable reason to exclude them. Outliers, in the strictest sense, are data points that are clearly erroneous.
The twin boys showed dramatic increases in dystrophin production
The entire hypothesis is that eteplirsen increases dystrophin production, which in turn, should increase ambulatory performance. Yet, these two boys showed marked increases in dystrophin-positive fibers at both weeks 24 (20.8% and 15.9%) and 48 (48.3% and 43.6%).
The significant increase in dystrophin production thus flies in the face of the researchers' decision to exclude them as "outliers" in the 6mwt. In fact, I would be much more bullish on eteplirsen's prospects if these extreme DMD patients showed stable ambulatory performance when taking eteplirsen, rather than getting worse. Biologically, these two boys were not outliers, and should never have been excluded from the analysis. In all fairness, the peer-reviewed paper did report the correct analysis, but the company has actively chosen to promote the more liberal interpretation of the results.
So let me get this straight. Dystrophin production increases yet ambulatory performance worsens?
Yes. This is why Sarepta is pushing the FDA to accept dystrophin production as their primary endpoint in an NDA filing, not the 6mwt that drisapersen was evaluated on. The FDA has yet to agree this is an appropriate primary endpoint, and I doubt they ever will.
Bottom line: Eteplirsen does not appear to provide a clinically meaningful benefit in terms of increased walking performance when the data is analyzed properly. To get the desired result, Sarepta had to inappropriately massage this miniscule dataset, and forgo any correction for Type I error.
But that's not all…
Why Sarepta chose such a small sample size is beyond me. Phase II clinical trials for drisapersen included 48 subjects and didn't involve any unwarranted data massaging. Sarepta's design, by contrast, is so small that Type I error corrections cannot be applied without losing a big chunk of statistical power, which is probably why it wasn't performed. The FDA tends to like to err on the side of caution, and it's common for them to ask for the more conservative p-values post-correction. My bet is that Sarepta's story falls apart at that point.
Finally, the sample size is also a major issue in terms of judging the safety profile of eteplirsen. Basically, you cannot say much at all. I'm not sure how important safety will be for such a devastating disease, but it further shows that the company is putting the cart before the horse.
Is Sarepta aware of these problems?
Sarepta is most definitely aware of these issues. They did report the correct analyses in their peer-reviewed work, and were upfront about not correcting for Type I error. It just wasn't trumpeted in their marketing materials to investors.
My guess is that Sarepta is deeply concerned that eteplirsen won't show clinical efficacy in a larger Phase III trial, and this is why they are pushing to get the drug approved now on the basis of a limited dataset. Remember, drisapersen showed very similar results in Phase II trials but failed dramatically in Phase III. And investors should be aware that Phase III failures are the rule, not the exception, for antisense oligonucleotide (AOs) drugs. Drisapersen's failure was only the latest in a very long list, and Sarepta is mostly certainly aware of this issue.
While Sarepta is now planning a "confirmatory" Phase III trial, one has to wonder why they don't simply wait until they have a robust data set. Their rather liberal stats showing clinical efficacy are intriguing at best, and certainly warrant further study. However, in light of the issues discussed above, I see no reason why the FDA would approve eteplirsen at this time. Given that AOs have a poor track record in Phase III trials, the FDA will more than likely exercise caution when reviewing eteplirsen's NDA in 2014.
How far will SRPT shares fall?
Given that most of SRPT's 60% rise recently has been due to the positive news surrounding eteplirsen, I see no good reason why the stock won't give back most, if not all, of those gains. Prosensa's Phase III bull run ended in a bloody 70% one day drop after drisapersen reported negative results, and is still bleeding to this day. As such, I think a 50% drop is well within reason upon the near certain FDA rejection of eteplirsen in 2014. Shorts appear to also share this bearish outlook, as they now hold 33% of the float. In conclusion, SRPT is an excellent short candidate with limited risk.
Disclosure: I am short SRPT. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.