I am a physician who has long followed antisense technology. I have been intrigued with the possibilities related to exon skipping since the early days; and of Dr. Wilton’s sentinel work from Australia as it relates to possible use in DMD. It was with interest then that I read last week the FDA comments related to eteplirsen; and witnessed the massive upheaval those caused. This letter to you is a response after my reviewing some of the objections that FDA raised to an early NDA filing for Sarepta for this analog. [This is a copy of a letter sent to ODE and DNP directors.]
Of first importance for me was to find the source(s) of that “recent data” used to suggest that DMD boys at baseline of 350m would be expected to be stable up to two years. After searching, I can only assume that the basis for that statement was Dr. McDonald’s recent analysis of the placebo group data from the atalauren study. In reviewing that paper (primarily Fig. 5), though it contributes significantly to natural history data in at least certain types of DMD, I was very surprised that you would consider it sufficient for such strong pronouncements as seen last week. You appear to have applied it very broadly—well outside the scope of the study population. The effect seems to be your having thus ‘thrown out’ ALL prior full deletion natural history data! After reviewing this paper and other natural history studies, I’m very concerned that you (FDA) have actually INVERTED the supportable conclusions from that data set! I will try to explain. [Other 2013 publications reviewed include Mazzone , and Henricson.]
McDonald’s atalauren data for most measures is divided into 7yr sections. However, in his Fig. 5 (the critical one for this point), he offers no similar data separation (i.e., in the 350m, no 7 plot). Nine of the 13 boys 350m group. I have communicated with Dr. MacDonald asking that he generate such a figure separating boys 7 and 350m and make that generally available.
Howsoever that may be, I extracted as much data as I could from figure 5 and reviewed the same. Some of the observations:
When the 350m is recalculated, (n=25) it shows a 6MWD value which is ~424m at baseline for the atalauren placebo subgroup. Significantly, this baseline mean is about as much greater than the eteplirsen mean as eteplirsen’s is greater than GSK-drisaperson’s mean! Thus, McDonald’s 7yr/ 350m group should be expected to perform stability-wise as much better than eteplirsen group, as eteplirsen should over drisaperson group. [Under the hypothesis that the ‘higher the baseline 6MWD, the greater the stability--until
Remarkable directional variability, rather than stability stands out to me in this 7yr./ 350m group. Most impressive is that 12 (of 25) boys DECLINED–with a mean decline of ~50m(-12%) OVER ONLY 48WK. With the known natural history of DMD, by 96 weeks, the majority, if not all, of these boys should have declined further (esp. see Mazzone, below). Twelve (12) boys showed some, but less increase: 48wk mean +19m (+4.5%). One boy didn’t show a detectable change in the Figure.
Also, I noted that there were NINE boys (of the 12 decliners in this ‘stable for up to 2 years’ group who had declines of 30m, whereas only THREE boys had INCREASES of 30m. That is 36% of that total population! Statistically stable? Perhaps. Clinically stable up to 2 years? Not likely. As a physician, it would worry me that 36% of the boys declined at significant levels. With FURTHER decline near-term probable to continue for most of these, it is likely that significant and possibly irreversible morbidity will occur within the next year in some/most of the boys. (again, Mazzone below).
Again, this group as a whole ( 7/ 350) had baseline mean of ~424m; AND OF THAT VERY HIGH FUNCTIONING GROUP 48% (n=12) DECLINED A MEAN OF ~5OM in LESS than 1 year. It seems that your statement last week about ‘stability’ being expected in the etep group; and such predicted ‘stability’ was based in large part on this group’s ‘stability’. In my reviewing the data, I worry that your statistical analysis inappropriately masks potentially serious clinical issues already under way. It implies there is ONE population, with a predictably homogenous clinical course to be expected. In fact, even this McDonald cohort is divergent AND already diverging. Statistical analysis which then assures stability well beyond the end of the study hides an already-identifiable subgroup at risk. I think it does that here; and thereby puts strains on the data which it isn’t strong enough to support.