Negative FDA decisions often seem left field to investors who...

Negative FDA decisions often seem left field to investors who have been told lots of reassuring statements about how“compelling” their trial results are.

And the FDA doesn’t like being the whipping boy when things go wrong … and so in not too many years the grounds for rejectionwill be made public and investors will be much better informed about what went wrong and why.

My reading of tea leaves would go something like as follows.

The key feature of confirmatory data is that it arises from pre-specified hypotheses. So the idea of retrospectively generating confirmatory data is a bit of an oxymoron.Typically one talks about a confirmatory trial inthe context of producing a statistically and clinically significant result on the primary outcome. And your primary outcome sets your indication.

In this case SPL have applied for two indications– the treatment of BV and the prevention of recurrent BV.  So you would guess that the FDA hasn’t foundthe P3 confirmatory data very compelling for one or both of these indications.

The confirmatory data for the treatment of BV presumably comes from the 2013 P3 trial. This trial failed on its primary endpoint. Only half the women were cured by the end of treatment and by 2-3weeks post treatment cure rates (28%) were no better than placebo.  Hardly compelling and because of the failure SPL changed tack and focussed on recurrent BV.

Recurrent BV is supported by the two 2017 P3 trials – in the US and Europe. In these trials women who had been successfully treated with an antibiotic (metronidazole) were recruited to test whether vivagel when used for 16 consecutive weeks was better than placebo for preventing a reoccurrence of BV.

The trials were conducted under a SPA. The results were statistically significant. In the US - vivagel 44% versus 54% placeboreoccurrence:  Europe - 16% vivagelversus 23% placebo.Clearly rates of reoccurrence are very dependenton who you recruit.  The benefit ofvivagel is very modest – around 10% or less over placebo.  

SPL claim there is a difference between “doing nothing” and receiving a placebo.  SPLargue that the true comparator is “doing nothing” and that the “doing nothing”rate can be derived from the literature and clinical judgment. SPL estimate itat around 65%. So it is claimed the “true” benefit from vivagel is larger than it might seem.

The key problem with this argument is that whyhave a placebo controlled trial in the first place if you want to use some other comparator. The fact that people have better outcomes in clinical trials even when on placebo is just bad luck for the active.

My hunch is that this difference of 10% betterthan placebo or the claimed 20%+ from doing nothing is pretty important. SPLdidn’t calculate this second rate for idle curiosity – 20% better is much morelikely to where you might think clinical significance lies. But note the “doing nothing” rate was notpre-specified – and so by using this as the comparator you lose theconfirmatory element of those two P3 trials. Its retrospective and so you areback to square one.

On this reading of the tea leaves its hard to seehow SPL thought they had a hope really. Of-course SPL have to act surprised anddisappointed. To do otherwise would be to admit that the “compelling” bit ofthe results was simply puffery.