FDA Hard-Pressed to Approve Biomarin, But Not Sarepta Drug

When Biomarin late last year bought Prosensa for its experimental exon skipper drisapersen for the treatment of Duchenne Muscular Dystrophy (for $680M plus potential milestones), it exuded confidence about the likelihood of getting approval for the 2’-O-methyl phosphorothioate antisense molecule.  This, despite of the fact that drisapersen failed in a pivotal phase III trial of 186 patients which prompted the old partner GSK to dump the drug and walk away.

Tenuous early evidence for drisapersen in earlier trials

The confidence is largely based on some supposedly successful earlier trials, especially a multi-center, randomized, blinded 53-patient phase II study which had seen improvements in the 6 minute walk distance (6MWD) at week 25, the primary endpoint of the study (Voit et al. 2014).  

This, however was statistically significant only the case in the subgroup of patients that received drisapersen continuously (à treatment in 10 out of 10 weeks with 6mg/kg), but not in patients which were treated identically, except for the small difference in skipping the last week in a 10-week treatment cycle.

At week 49, the difference with placebo failed to reach statistical significance and Prosensa had to resort to pooling both subgroups to claim victory for that time point. Similarly, drisapersen failed in obtaining statistically significant outcomes for other muscle function endpoints.

Since the mechanism of action for the DMD exon skipping candidates is to change splicing of the mutated dystrophin transcript to a form in which the reading frame is restored with recovery of partial activity, it is important to understand the relationship between drug treatment and dystrophin production.

Here, too, the evidence was less than robust.  For example, even when applying the sensitive immunofluorescence technique, no increase or even a decrease in dystrophin was seen in almost half of treated subjects.  With the less sensitive Western blot, an increase in dystrophin was seen in only a third of treated subjects (0 for placebo).

Therefore, given the failed phase III trial and the less than robust earlier evidence in favor of the drug, I struggle to understand Biomarin’s confidence in obtaining approval in 2016.   

The importance of dystrophin as a surrogate endpoint

Part of the difficulty of obtaining statistically significant results for muscle function endpoints is most likely due to the small patient size (orphan disease affecting ~1 in 3500 male births) and the consequent need to pool boys at various stages of the disease together in a given trial.  It would thus not be surprising if say obtaining 10% levels of normal or Becker-type dystrophin will translate into very meaningful clinical benefit in some, but not other boys.

This will be an even more challenging problem for the DMD subgroups that are not amenable to exon 51-based exon skipping which is targeted by drisapersen.  Probably insurmountable for first-generation chemistries like drisapersen.

Accordingly, in both the drisapersen and the competitive PMO-based eteplirsen trials, it has not been possible to correlate dystrophin production with functional outcomes.

For that reason, I strongly support the importance of establishing reliable, quantitative methods to measure dystrophin in clinical trials (there was an FDA workshop related to this last week).  Dystrophin-dependent markers may also be acceptable if they can be measured by means that do not involve taking painful muscle biopsies.  For example, serum-based microRNAs as developed by Rosetta Genomics and Marina Biotech would be of interest here.

Eteplirsen before drisapersen

I thus find it difficult to grasp the notion of rejecting the current crop of exon skippers like drisapersen or eteplirsen should they be found to produce functional dystrophin with few side effects.  After all, it is the loss of dystrophin function that causes Duchenne Muscular Dystrophy and one has to wonder how generating additional dystrophin cannot be beneficial to patients, especially since the principle behind drisapersen and eteplirsen is strongly supported by human genetic evidence (à Becker’s Muscular Dystrophy).

In this world, it has got to be eteplirsen that should be first in line for regulatory approval.  This is because there is overwhelming evidence (e.g. Heemskerk et al., 2009; Sarepta's Barclays presentation March 12, 2015) that the PMO-based drug is much more potent than drisapersen which, let’s face it, is based on stone-age antisense chemistry (2’-O-methyl phosphorothioate).  Such chemistry is characterized by minimal efficacy and dose-limiting toxicities, especially renal in the case of drisapersen.

In a paper comparing 2’-O-methyl to PMO chemistry for DMD exon skipping conducted by researchers close to eteplirsen, it was found that at same doses in mice, PMO chemistry is moderately to vastly more potent than 2’-O-methyl phosphorothioate antisense compounds of a size comparable to drisapersen.  The extent of the difference depended on whether the human or mouse dystrophin were targeted and the target sequence.   Unsurprisingly given the acrimonious competition between the two parties, Sarepta has also picked up on this and continued along these lines by showing that in addition to chemistry, eteplirsen has the edge over drisapersen in terms of the targeted sequence:

Sure, there is the theoretical caveat that PMO and 2’-O-methyl scale differently from mice to humans and that what is the most potent target sequence for one chemistry does not necessarily have to be the most potent one for the other.  Intuitively, however, the differences are too big for these factors to compensate the preclinical evidence.  Also, keep in mind that in the clinic, eteplirsen is being given at 5 to almost 10-fold increased doses than drisapersen and, on top of that, is much safer and better tolerated than drisapersen.

Because of this and the competition, it is not surprising and disingenuous when Biomarin would now suddenly like to de-emphasize the importance of dystrophin as a surrogate biomarker (see last week's workshop).

Dear regulatory agency, if you approve drisapersen, you cannot deny eteplirsen.  Sure, drisapersen has been tested in more patients than eteplirsen and Sarepta has conducted a clinical trial in the worst possible manner and probably ‘embellished’/overstated some of their results, including the dystrophin evidence.  However, given that eteplirsen almost certainly generates more dystrophin than drisapersen, the highly favorable side effect profile of eteplirsen (also in comparison to drisapersen), and in light of the 6MWD issue that applies to both drug candidates, the question is whether the bureaucratic application of rules should trump scientific evidence and patient interests.

Disclosure: I am long SRPT based on the notion that Biomarin, with its orphan disease savvy, will turn out to be the biggest supporter of eteplirsen getting approval this time around.  Additionally, the agency is partly responsible for the long duration of the ongoing eteplirsen trial (close to 4 years soon) and the repeated taking of muscle biopsies, and after all this taking away hope from patients and their close ones is difficult to fathom.

Sarepta, Biomarin Move Over- Here Come tcDNAs

In a head-to-head comparative study in NatureMedicine, Goyenvalle and colleagues claim that tricycloDNAs (tcDNAs), a relatively unexplored constrained nucleic acid analogue commercialized by SYNTHENA have superior efficacy over 2’-O-methyl oligos and PMOs for therapeutic splice modulation in mouse models of Duchenne Muscular Dystrophy (DMD). 

In addition to ~3 to 5-fold enhancements in dystrophin exon skipping in various muscles compared to the competing chemistries, the authors report that only with tcDNAs there was splice modulation also for the dystrophin isoform expressed in the brain.  This was accompanied by improvements in the behavioral and cognitive abnormalities in this model.   Apparently, these neurological defects are another important treatment goal in DMD.

Splice modulation was accompanied by ~2 and 20ug/g oligo concentrations in brain and muscle, respectively, following the administration of relatively large weekly doses of 200mg/kg that were necessary particularly for the correction of the neurological symptoms.

Interestingly, such delivery and target modulation was achieved without encapsulation or further modification technologies.  Whereas a 2012 publication by some of the same authors in collaboration with ISIS Pharmaceuticals explored the use of tcDNAs for RNaseH-mediated gene knockdown in conjunction with the phosphorothioate backbone known to greatly improve biodistribution and cell uptake, tcDNAs here were applied in their naked form.

The authors attribute this to an apparently spontaneous tendency of tcDNAs to self-assemble into ~100nm nanoparticles.   

The results seem to warrant the development of tcDNA for DMD exon skipping.  Critical to their success in the clinic will be the safety and tolerability of tcDNAs in Man with the kidney predicted to be the dose-limiting organ.  

The study is also a reminder that there are now a number of other chemistries, including the high-affinity CRN chemistries by ISIS Pharmaceuticals and Marina Biotech, the latter with a stated focus on DMD, which similarly promise improvements over the trail-blazing 2’-O-methyl and PMO chemistries.  

Sarepta Ditches Dystrophin Assay Getting Closest to Being Quantitative

I am aware that my interpretations of Sarepta’s exon-skipping data for Duchenne Muscular Dystrophy are not universally embraced.  But one point that I believe all of us can agree on is that it would be important to determine whether the antisense oligo eteplirsen can restore sufficient Becker-type dystrophin to have a therapeutic effect.  Although falling short of a controlled clinical outcome study, one way to gain accelerated approval for this devastating disease would be to first establish what this ‘magic’ level is through historical outcomes studies and then compare this value to the levels actually restored following drug treatment. 

Western blot not worth the film

But when you listened to last week’s presentation by Sarepta CEO Chris Garabedian (link to transcript), the company appears to have dropped the idea of demonstrating such altogether by declaring dystrophin Western blots as unreliable.  I agree that Western blots do not lend themselves to quantitation.  Moreover, I have sharply criticized a number of technical aspects with the eteplirsen Western blot: spliced gels, uneven loading controls, cropped band of unknown identity shown.  

The latter point is particularly bothersome as any molecular biologist would know that even two supposedly identically processed protein samples will show numerous bands that are present in just one of the samples if only you expose the Westerns for long enough.  Ergo, the particular Western blot does not live up to being even merely ‘supportive’ as the company had claimed until recently because it now seems that even the company is only guessing that the cropped band shown is the one based on predicted size (note: proteins do not even migrate on Western gels according to known size...charge etc).

All this begs the question of why Sarepta never developed a reliable quantitative assay (e.g. an ELISA) before entering clinical development and making the case for biomarker-based accelerated approval.

Counting dystropin-positive fibers by immunofluorescence does not address magic number

This alas leaves us with the immunofluorescence (IF) assay data.  Sarepta and their collaborator have counted and presented numbers, so this surely must mean that the IF data are ‘quantitative’?

Wrong.  

As I keep pressing the point, the IF assay (at best) can only determine if there was a increase in dystrophin.  Manipulate the exposure times and picture contrasts enough and you can magnify any difference in dystrophin levels out of proportion. Again, this is no revelation to the molecular biologists and hobby photographers here, and I won’t even start to discuss the issues arising out of comparing IF numbers from experiments conducted at different dates and from different muscles.

By way of example, say you obtain an average signal level of 1.5 due to naturally occurring alternative splicing noise when measuring background, with the vast majority of fibers being below 2.0 and with some rare revertant fibers exceeding that value manifold.  Then you treat with eteplirsen and you find that the signal level increased to between 2.2 and 2.8 in 70% of fibers.  You then set the background arbitrarily to be 2.0, subtract the few revertant fibers, et voila, you determine that 65-70% of fibers have started to express dystrophin following drug treatment although the absolute increase in dystrophin produced is well below what we would predict to be therapeutic: from 1.5 to about 2.5. 

    

Leaving it open what that ‘magic number’ actually refers to

Consequently, IF is ill suited to address the issue of the ‘magic number’ as Mr Garabedian called it.  It would have been incredibly helpful for him to have elaborated what he actually meant with the ‘magic number’.  Does he really believe that counting fibers allows you to get at this issue, as he implied in his answer*, or would he agree that it should be with respect to the absolute amount of dystrophin?  I hope the company will clarify that point almost as quickly as they declared the image duplication event an honest mistake.

What do you think?  Should the ‘magic number’ refer to counting dystrophin-positive fibers or quantitating the absolute amount of dystrophin?  Participate on the poll on the right hand top.

* Making a circular argument, the CEO said that ‘yes’ they know they have surpassed the magic number because of the claimed 6MWT improvements.  Note that in making the case for a biomarker like dystrophin, it should be the amount of biomarker that supports functional outcomes like 6MWT e.g. through some quantitative correlation- not the other way around. 

FDA Questions around Dystrophin Quantitation by Sarepta Delay DMD Drug

As most of you know, yesterday was the day that morpholino antisense company Sarepta would provide guidance on whether they would be seeking accelerated approval for the Duchenne Muscular Dystrophy (DMD) exon skipping drug eteplirsen based on their interactions with the FDA.  This was a widely watched event not just for eteplirsen and boys with DMD, but also because regulators highly supportive of Sarepta’s meager data package would indicate a new lowered standard for the approval of rare, severe diseases.

Dystrophin as surrogate biomarker

Accelerated approval may be granted by the FDA in the absence of full clinical evidence of drug efficacy if the results from surrogate biomarkers that are thought to reasonably predict clinical outcome pointed in the right direction.  For dystrophin exon skipping this is an important potential avenue given that for most genetic lesions causing the disease it would be difficult to conduct full-fledged placebo-controlled clinical trials.

Although this is not really the case for exon 51 candidates, the most frequently affected exon in DMD and that eteplirsen is addressing, Sarepta ran one of the worst designed and conducted clinical trials for this indication involving just 12 boys (in case you are asking again why I hold this view: minute size of study when larger patient numbers were clearly feasible as the Prosensa/GSK competition shows; single-center; key supportive data miraculously collected with unblinding of the trial; key secondary endpoint- 6MWT- added with unblinding of trial; exclusion of patients in results analysis as company saw fit etc).  As a result, obtaining approval by simply showing dystrophin production has been Sarepta’s best hope with the data at hand.

Hardly anyone would argue that dystrophin was an inadequate surrogate marker.  The loss/absence of dystrophin is what causes the muscle wasting, and there is strong evidence that restoring dystrophin or even a somewhat less functional Becker’s type truncated version would be beneficial.  It would also be no surprise to see some increase in dystrophin with a gain-of-function approach such as this exon skipping one. 

No, the key questions are how much of Becker’s type dystrophin production was needed to be clinically meaningful to patients and how you actually quantitatively measure the increase in such dystrophin.

Questionable science

I don’t want to argue here about the amount of Becker’s type dystrophin needed to be clinically meaningful.  The reason is that if you do not have reliable numbers, such a question is moot.  As I had explained in a post a few months ago, the data presentation by Sarepta on the amount dystrophin produced reminded me of some of the worse cases of ‘data management’ and misinterpretation that you sometimes come across in the literature: there was apparently unequal loadings between drug-treated and control samples which suggested a high amount of skipped dystrophin by Western blot, a molecular technique looking at protein content; the RT-PCR that was supposed to support meaningful skipped dystrophin production by looking at the RNA (not protein) level actually suggested the opposite as the PCR cycle number seemed inappropriately high. 

Moreover, choosing percent dystrophin positive fibers instead of the absolute amount of dystrophin produced as the primary endpoint in the clinical trial always seemed wrong to me as for this gain-of-function approach you could imagine a very low amount of dystrophin that can be detected by the particular assay could yield very high percent positive fiber numbers, but of which the clinical relevance would still be highly uncertain (e.g. if sensitivity of assay could pick up 1% of normal, and the restored dystrophin was homogeneously distributed between the fibers).

And these are only the obvious questions.  What I would also like to know (and the FDA may or may not know) following the piecemeal data release by the company and the surprising finding that after 12 weeks no dystrophin restoration could be observed, yet after 24 weeks the company apparently could, would be the comparability of the data (e.g. the immunohistochemistry/IHC) across the different time-points, who performed the analysis, the appropriateness of the antibody used in the IHC and the Western blots, the collection method of the biopsies etc etc

FDA also doubtful

It is becoming clearer with each report on the company’s interaction with the agency that this methods issue is a real sticking point before a proper evaluation can be conducted.  This can also be seen from today’s press release: ‘The Agency, however, requested additional information related to the methodology and verification of dystrophin quantification.’

Alas, this should come as no surprise.  It is still good to see confirmation that the FDA is taking a scientifically sound approach, but as the discussion around eteplirsen had become quite politicized, one could have been excused for falling for the belief (as I’m sure countless retail shareholders did) that politicians and not scientists have taken over the eteplirsen approval process.  I had even considered whether we were about to witness a landmark event lowering the evidence bar for the approval of orphan drugs to unprecedentedly low levels.  Apparently not.

PS: The fact that Sarepta would decide to file for accelerated approval sometimes in the first half of 2014 should not come as a surprising.  Even if they themselves had considerable doubts about the success of accelerated approval or even about acceptance for review following their FDA interactions, there was nothing to be gained from not stating such intent: the stock price would have plummeted much more than the roughly 20% it did yesterday.  The interim now can be used to exercise their ATM (=issuing new shares on the actively trading markets) at still much elevated levels.

Disclaimer: The above was written based to the best of my knowledge and belief. Nevertheless, no guarantees can be given as to the factual accuracy of the content. I should also add that I am not 'against' eteplirsen.  I just don't think that there is sufficient evidence yet to decide either way. And for parents eager to get their kids on eteplirsen now, I believe, given the apparent safety of the drug, that a way could be found in which Sarepta were to reciprocate the support by the patient community and provide the drug at cost until a possible full approval.