Chalk up a fail for Nuplazid, Acadia’s spin notwithstanding

As one headache for Acadia eases another begins, with unconvincing results from a mid-stage depression trial of Nuplazid.

The nightmare scenario of US withdrawal of the antipsychotic drug Nuplazid having recently been lifted, Acadia was today looking for promising signs from a mid-stage depression trial to make the case to go into phase III and give the drug a second string to its bow.

However, closer inspection of the study’s design and the data revealed make this look like another case of a failed CNS trial being spun positively. Acadia has touted the result as statistically significant on its primary endpoint, but the data raise several red flags, and it is hard to discern precisely what Nuplazid is demonstrating.

The markets reacted sceptically, Acadia stock climbing 16% in the premarket as the nominally positive headline numbers hit, before falling 6% in early trade.

Lack of Clarity

The depression trial in question, called Clarity, used a sequential parallel comparison design, it was revealed today. This was intended to eliminate the confounding effect of subjects who responded to placebo: placebo non-responders in Clarity’s first five-week stage were re-randomised to Nuplazid or placebo for another five weeks.

The primary efficacy measure was HAMD-17 total score, and of the 207 initial enrollees (only 203 appear in the final analysis) 58 were re-randomised for stage two. All subjects had to have failed first-line SSRI or SNRI antidepressants, and all remained on this background medication in addition to Nuplazid or placebo.

The problems begin when it becomes apparent that for Acadia to declare the result a win it has had to pool the separate datasets from stage one and two of Clarity. Until now analysts had only spoken of change in HAMD-17 between week one and 10, and this is also what Clarity’s entry suggests.

Whether Acadia changed the precise primary endpoint criteria before unblinding Clarity is not obvious, but the fact is that the study seems to have achieved the opposite of what it had intended: having enriched the dataset for placebo non-responders it then failed to show an effect in these subjects.

This is perhaps Clarity’s most important take-home message. The study’s first stage showed a strong treatment effect favouring Nuplazid, but stage two was negative, to the extent of numerically favouring placebo.

And the highly positive effect in stage one might be explained by a patient imbalance: subjects who initially got Nuplazid had been depressed for 37 months on average, versus 25 months for those on placebo, and thus were possibly more likely to respond.

Either way, stage one drove the pooled analysis, which yielded a somewhat unconvincing p value of 0.039.

Summary of results from the Clarity trial
  Nuplazid Placebo
Stage 1 (weeks 1 to 5)
Patient number* 51 152
Change in HAMD-17 total score -11.5 -7.5
p value 0.0003
Stage 2 (non-responders to placebo in stage 1; weeks 6-10)
Patient number 29 29
Change in HAMD-17 total score -2.8 -3.3
p value Not significant
Pooled stages 1 & 2
Patient number* 51 152
Change in HAMD-17 total score Not disclosed Not disclosed
p value 0.039
Source: Acadia presentation. Note: *actual number enrolled was 207.

On an analyst call Acadia argued that the results from stage two were hard to interpret. It claimed to have “overselected for difficult-to-treat patients not likely to respond” to either placebo or active treatment, and said only half of the assumed number of patients were actually re-randomised for stage two.

There had earlier been concerns that Clarity’s relatively small size would have left it insufficiently powered to show significance, but this does not seem to have been a factor. A small study also says little about safety; Nuplazid in its approved indication of Parkinson’s disease psychosis has faced safety concerns, though fears that it would be withdrawn were recently allayed (Acadia gets Nuplazid boost, but doubts linger, September 21, 2018).

However unclear a signal Clarity has generated, some analysts fear something entirely different. Stifel, for instance, pointed to commercial doubts: Nuplazid costs twice as much as Otsuka’s antipsychotic Rexulti, meaning that it might be used after this drug and Abilify, at the same time as facing emerging depression treatments from Sage, Johnson & Johnson and Allergan that “may be cheaper”.

Acadia said it expected Clarity to serve as one of two pivotal studies required for approval, but nevertheless said it would run two phase IIIs. The design of these will clearly be of vital importance given the confusing Clarity dataset.

Ironically, Nuplazid might be showing hints of efficacy, and in hindsight a standard rather than sequential parallel comparison study might have shown this more clearly. Sceptics will nevertheless note that marginal phase II efficacy usually disappears in the more robust setting of phase III.

Share This Article