Sarepta Discloses Dystrophin Expression Data at World Muscle Society Congress

Last week, I criticized Sarepta Therapeutics for failing to disclose critical dystrophin expression data regarding their exon-skipping drug candidate eteplirsen for the treatment of DMD.  Instead, they merely disclosed the percentage of muscle fibers staining positive for dystrophin, i.e. fibers that expressed some dystrophin, which, of course, would be a much more impressively sounding number.  Adding to what I perceived as dubious data presentation was 6-minute-walk test (6MWT) data of which the integrity seemed compromised based on cherry-picking the patients and the fact that the apparent improvement occurred in the open-label extension phase of the study.

Admittedly, the percentage positive fiber number was a previously defined primary endpoint whereas the amount of overall dystrophin relative to normal/healthy was not.  Therefore, like other biotechs frequently do, it might always have been Sarepta’s intention that only the primary endpoint data be presented initially ('top-line data'), with the balance to be disclosed at a future scientific conference or peer-reviewed presentation.

Accordingly, Sarepta today presented additional data at the World Muscle Society Confress in Perth, Australia.  I was pleased to see that at least some dystrophin expression Western blot data were revealed (slide 12)- with a direct comparison to ‘normal’ at that.

First of all, you can see on the left the ‘normal control’ (healthy) comparison.  The labeling seems to be somewhat off as I believe the outermost left lane represents the sample.  You can also see that the normalization control ‘actin’ was less abundant in the control sample compared to the patient samples (by naked eye about 3-5 fold).  This means that the 'normal' dystrophin signal is an under-estimate relative to the patient samples. What is more, because the dystrophin signal in the ‘normal control’ is over-saturated (big black blotch), the ‘real’ relative dystrophin amount should be even more compared to the patients'.  Because there is no standard curve and because of the apparent loading differences (à actin), it becomes an art to estimate the relative dystrophin amount restored by eteplirsen treatment, but with 7 years of doing the same type of experiments at the bench under my belt, I would guesstimate that the true value is about 2-5% of normal.  Certainly not the 20% believed to be necessary to start impacting the disease phenotype. [Correction 10/13/2012: This was an inadvertent mis-characterization of the quoted source, please see the discussion in the Comments section].

Supporting the notion that the Western Blot overestimates restored dystrophin levels, is the RT-PCR experiment on the right.  Again, the larger band which represents the mutated transcript is much more abundant than the re-spliced signal below, indicating that only a minor fraction of the mutant transcripts were re-spliced.  Moreover, smaller PCR products amplify more efficiently compared to larger ones, meaning that the apparent splice correction is an over-estimate.  Similarly, because the mutated and ‘normal’ PCR signals are almost the same (you would expect an mRNA with premature stop codons to be destabilized relative to wild-type), it is likely that the normally spliced bands have reached saturation, further over-estimating the splice correction in the RT-PCR experiment.

Overall, whatever Sarepta’s motives were behind failing to disclose the absolute dystrophin data last week, they clearly provided some of the necessary transparency at the WMSC presentation.  Having said that, I believe the data support the underlying concern that the splice correction may not be enough to be therapeutic.  I would not exclude it, however, as the 20% value, of course, is also an estimate.  Blame a less than rigorous trial design for not knowing for sure for quite some time to come. 

Sarepta Fails to Report Meaningful Dystrophin Expression Data, Falls Behind Prosensa

Restoring truncated, yet largely functional dystrophin in Duchenne Muscular Dystrophy (DMD) by exon skipping is a promising approach to this devastating disease.  For this, however, to have any meaningful therapeutic impact, it has been estimated that one would have to achieve about 20% of wild-type (probably somewhat higher for truncated versions) dystrophin levels in all muscle fibers.

Releasing 48-week data from one of the most manipulated clinical trials that I have seen, Sarepta Therapeutics just reported that treatment with their exon-skipping antisense morpholino eteplirsen resulted in 34-52% dystrophin-positive muscle fibers.  Impressive? I can’t tell, but what is more worrisome is that I don’t know why the company does not tell either.

Percent muscle fibers expressing dystrophin does not mean equal amount of dystrophin restored.

In a New England Journal of Medicine paper last year, Sarepta’s competitor Prosensa (with partner GSK) reported in a similar-sized study that in 10 out of 12 patients, the percent of fibers expressing dystrophin following treatment with their exon skipper PRO051 was between 60 and 100%.  Yes, 60-100% and nost just 34-52% as in Sarepta's case.

What is more, the immunofluorescence intensity and Western blot data in the Prosensa study showed that those fibers expressing dystrophin did so at much reduced levels compared to healthy muscle, on the order of 5-20% (overall bulk levels).  In other words, it is quite likely that the ~50% of fibers that express dystrophin in the Sarepta study together express at most 3-10% of normal dystrophin, well short of what is expected to be therapeutic.  It is also possible, however, that Sarepta’s assay is very sensitive such that even fibers with just 1% or so of wild-type signals were counted in. 

Given that stating the percent fibers expressing any dystrophin is almost meaningless, you have to wonder why Sarepta has not presented the data.

But then you ask, then why did the lame start to walk in the study (à 6 minute-walk-test data)?  Actually, I don’t believe that this was shown in the study either.  Just because this is a devastating disease in children does not mean that rigorous trial design and execution can be entirely done away with.  Yet, in this already tiny 12-patient, single-center (!), and open-label (!) study, the company wisely discarded the worst responders early on allowing the 6 MWT numbers to be 'statistically significant'.   There seemed to be more subgroup analyses presented than patients in the study.  If I were a regulatory agency, I would strongly consider refusing to even just look at the 6 MWT data.

Why am I writing this?  First of all, I am obviously trying to find any shares to short.  And second of all,  what really irks me is that as orphan drugs become more and more popular in drug development, a number of companies are doing away with rigorous science, drive aggressive patient recruiting and disease awareness campaigns (best if a pediatric indication is involved), and expect the rest of us to subsidize the enormous costs of these drugs.  Fine, if the drugs worked, but not for placebos.  Wait another 5-10 years and there will be an enormous backlash and those that need access to these drugs will suffer the most.