The Thing That’s Bugging Me About Sarepta Therapeutics

So there are a few disclaimers in this post that I need to get out of the way right off the bat.   I am very critical of people who pretend to have expertise in subjects which they then go on to demonstrate that they clearly lack expertise in.   There’s a word for these people: charlatans.    Let me be clear: I am no sort of expert in biotech stocks.   I’m not even an *almost* expert in biotech stocks.  If you called me an idiot when it comes to biotech stocks I wouldn’t argue with you.   Now I’m going to write about a biotech stock.

The second disclaimer is that this post isn’t meant to be a background of what’s been going on with Sarepta Therapeutics ($SRPT – long), or a prediction of what might happen in the future.  I’m certainly not trying to convince anyone to buy or sell the stock.  This is really more of a post about math and logic than it is about Sarpeta or biotechnology.    Now let’s get to it.

Sarepta’s news today was that the FDA seems to have again done (another!) 180 degree about-face on how they feel about Sarepta filing a New Drug Application for Eteplirsen:  their drug to treat Duchenne Muscular Dystrophy (DMD).   $SRPT bulls (obviously) think that the drug works, as evidenced by the 6 minute walk test (6MWT) data compared to the “natural history” data for the disease – combined data from other studies of DMD and from other patients who have suffered from the debilitating and then lethal disease.    Perhaps the best simple summary of this data is from $SRPT’s recent presentation at the 19th Congress of the World Muscle Society.  If you click on that link, and the click on the first presentation, you’ll get a pdf that has the “money” slides.   The first important slide is the chart (pg 8) of the natural history of 6MWT progression in DMD boys.

page 8 - DMD 6mwt Natural History Data

page 8 – DMD 6mwt Natural History Data


Page 9 shows the 6MWT charts for SRPT’s two subgroups – their Eteplirsen cohort, and their placebo/delayed cohort who started out on a placebo and switched over to Eteplirsen after 24 weeks.



You can see that SRPT shades a grey area that they highlight as relevant, as their data shows dystrophin present in the patients starting roughly 12 weeks after treatment.   A dumbed down interpretation of those grey boxes is “it takes at least 12 weeks for the drug to start working.”   We’re going to ignore for a moment the FDA’s comments on the measurement of dystrophin which played a part in decimating the stock today.

Here’s the problem:  as those familiar with $SRPT are already aware, Sarepta modified their study to exclude 2 patients who deteriorated quickly and were unable to perform the 6MWT after 24 weeks.    Now look at the bottom of  the chart in slide 8 above: note the important part: “Natural history studies imputed zero values for patients who lost ambulation during follow up periods.”  In other words, Sarepta threw out their study’s zeroes, while the natural history data averaged them in.   Is that a problem?

Well, on the one hand, I already acknowledged above how, in some sense, it makes sense to exclude the data for the 2 boys who deteriorated before the drug had a chance to work.   The point of Sarepta’s study is to evaluate if the drug works.   The drug works by producing dystrophin, and these boys deteriorated before the drug could get their dystrophin levels up.   That’s not crazy, fraudulent, shady, manipulative, or any of the other adjectives some critics of the company would use to describe their data.

HOWEVER – once you start excluding data from the study, especially given the tiny sample size of Sarepta’s study, you have a problem if you then continue to compare $SRPT’s modified 6MWT data set to the un-modified natural history data set.   Do you see why?   I’m not a biotech expert, but I’m pretty good at math, logic, and statistics, and the thing bothering me here is that we’re doing an apples to oranges comparison.  You can manipulate SRPT’s data yourself to impute zero value 6MWT for the 2 boys who were excluded in the modified population, and you’ll get a very different picture from the one on page 9.

Kid Dynamite,” you’re saying, “don’t be an idiot – you already said TWICE that it seems reasonable to exclude those boys, as the drug didn’t have a chance to work on them”  Ah yes – indeed – but then we need to compare the SRPT data to *comparable* natural history data – for example: adjust all the natural history data so that we exclude all boys who stopped walking in the first 24 weeks instead of averaging their zeroes into the data set.      THAT would be an apples to apples comparison that would result in some very different looking line slopes on that natural history data chart.     Is it a coincidence that in today’s SEC filing Sarepta noted that the:

“FDA strongly advises the sponsor to obtain and submit patient-level natural history data. FDA is prepared to appeal to the academic groups holding the data to allow the sponsor a means to acquire the data.”

Look – Sarepta has a “small N” issue with their data set.   Everyone knows that.   Bulls cite comparisons to natural history data, but I think those comparisons are flawed.     Bears cite unreliability of small samples, especially given how Sarepta has chosen to massage the data and the patient populations.   However, even if one backs back into comparable 6MWT data by imputing zeroes for the 2 SRPT boys who lost ambulation, the Sarepta 6MWT slopes are still interesting compared to the presented natural history data.      They still  stabilize (albeit at a lower level, once the zeroes are included), and thus far haven’t shown much deterioration compared to natural history.   Bears would probably point out that this is the small N problem again, and that with a 10 person sample, anything can happen.    Bulls would cite the fact that with this debilitating disease, stability is virtually unheard of.

I have been involved with the Sarepta story for almost 2 years now.  That most certainly does not make me an expert on the stock, on biotech, on Duchenne Muscular Dystrophy, on scientific methods in FDA trials, or on FDA panel decisions.   I do, however, feel equipped to point out the flaw in a common data comparison, as noted above, regardless of the fact my conclusion doesn’t allow me to have any confidence making statements about the efficacy or lack thereof of Eteplirsen.

Rather than make inane non-expert statements of how I interpret the Eteplirsen data, I’ll simply try to repeat and synthesize my rambling thoughts above:  I would like to compare Sarepta’s small sample of data to individual data from the 6MWT natural history data set, rather than to averages which get bastardized by individual results.

If any readers have insight into the MMRM (Mixed Model Repeated Measures) statistical analysis Sarepta uses, I’d love to hear anything you know about it.   It surprises me that they are getting such significant p-values (in their comparisons to their own placebo/delayed cohort) from such small and variable data sets.


note: I am fully aware that I didn’t talk about safety, etc..   I am also completely uninterested in having a debate with those who shout “THE DRUG WORKS” without responding to the points I have raised above.


About Duchenne Muscular Dystrophy – from Sarepta’s own presentations:

“DMD is an X-linked rare degenerative neuromuscular disorder causing severe progressive muscle loss and premature death. DMD affects approximately one in every 3,500 boys born worldwide. A devastating and incurable muscle-wasting disease, DMD is associated with specific errors in the gene that codes for dystrophin, a protein that plays a key structural role in muscle fiber function. Progressive muscle weakness in the lower limbs spreads to the arms, neck and other areas. Eventually, increasing difficulty in breathing due to respiratory muscle dysfunction requires ventilation support, and cardiac dysfunction can lead to heart failure. The condition is universally fatal, and death usually occurs before the age of 30.”


Kid Dynamite is a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for sites to earn advertising fees by advertising and linking to If you click on my links and buy anything, even something other than the product advertised, I earn a small commission, yet you don't pay any extra. Thank you for your support.

The information in this blog post represents my own opinions and does not contain a recommendation for any particular security or investment. I or my affiliates may hold positions or other interests in securities mentioned in the Blog, please see my Disclaimer page for my full disclaimer.

blog comments powered by Disqus
Kiddynamitesworld Blog