<<Neuro, doesn't this look doable? >>
Before you get too excited, this is fairly old data, reported in December 2007. The events thereafter were interesting: Here is the sequence as was summarized in NI February 2008:
<<Epix's Epic Embarrassment
"Michael G. Kauffman, M.D., Ph.D., chief executive officer
of EPIX Pharmaceuticals stated, “We are very excited
by the measurable impact on memory and cognition,
achieved in such a short period of time, in a trial that was
designed primarily to assess safety and tolerability."
--Press release from Epix Pharmaceuticals, 12/18/07
"When things seem to be too good to be true, they usually
are (too good to be true): Epix's Alzheimer's trial for its
5HT-4 drug PRX-03140 showed an amazing 5.7 point
ADAS-cog improvement in just two weeks...albeit in just
one 10 patient dose cohort, with two 14 point responders,
the mean change for the others was 3.3 points, and the
other six cohorts did not show significant benefit. And
there was no benefit if the patients also received Aricept-
-Epix states that these patients hadn't been washed out,
and thus were already at the cholinergic ceiling (dubious
trial design and execution). But other than that, these
were revolutionary results. So revolutionary, in fact,
that NI strongly doubts that they will be replicated.
Because they are not only revolutionary, they make no
sense."
--NeuroInvestment 1/06/08
"The updated results described below reflect the
correction of previously undetected errors that were
included in the trial results as provided to the company
from a third party contract research organization (CRO)
and as reported by the company in a recent press release,
as well as newly available data on other measures of
cognition.
As a result of errors made in the transcription of data and
calculation of the Alzheimer’s Disease Assessment Scale
cognitive subscale (ADAS-cog) score, an independent reanalysis
of the data has been conducted. The corrected
results show that patients receiving 150 mg of PRX-03140
orally once daily as monotherapy achieved a mean 3.6
point improvement on the ADAS-cog versus a 0.9 point
worsening in patients on placebo, which continues to be
statistically significant (p= 0.021).
--Press release from Epix Pharmaceuticals, 1/15/08
"Oh. That's very different. Never mind."
--Gilda Radner 1976
Perhaps in the future, if circumstances warrant, Epix will
conduct the "independent re-analysis' of suspect data
before broadcasting them. (NI February 08)>>
Basically, so far as I can tell, no one takes even the corrected data all that seriously. Between statistical incompetence and flukiness, this is not something that anyone should aim to emulate.
BTW--the CEO was subsequently fired. Replaced by interim CEO Elkan Gamzu, who I know well from Cambridge Neuroscience and Pharmos. And now they are presenting 16 month old data for lack of anything else to say.
NeuroInvestment
AI
