Dear Stephen

My question is,
>I have so far seen very few drug studies with fMRI(as compared to PET). Is
>there any intrinsic reason why fMRI is less suited to drug effects
>studies? 

For what it's worth, we are presenting a poster on a psychopharmacological
fMRI study at the Cognitive Neuroscience meeting in San Francisco in 3
weeks.  This was similar to your design in that we were were interested in
the differential effects of the drugs on a task versus a control state i.e.
a cognitive rather than a pharmacological question (which I think many of
the fMRI studies so far have been).  We gave 12 subjects placebo, clonidine
or guanfacine in a within-Ss design (each subject scanned on 3 different
sessions with a different treatment each time) and examined the
differential effects of the drugs on spatial and temporal attentional
orienting compared to a control task.  Using a random effects analysis, we
found clearly differential (and sensible from a theoretical point of view)
effects of the drugs on the task versus control states.  Personally, I was
so pleased with the results that I have forsaken PET, and am now planning
to do all of my psychopharmacological studies with fMRI.

I think one of the reasons these sorts of studies are so rare is that there
were some serious considerations regarding the statistical viability of
comparing between separate treatment groups - however, as mentioned in a
previous posting, the random effects analysis has taken care of this.


>would very much appreciate some pointers. I realize that for fMRI one
>would need to treat the drug effect as an interaction effect (i.e., how
>does the drug treatment affect the difference between the activation
>memory task and a reference control task?)

In fact, this is true for psychopharmacological PET studies of cognitive
function as well.  Without a reference state you have no way of knowing
whether the observed effects of the drug (relative to placebo) during your
cognitive task are specific to the experimental cognitive state or merely
reflect more general changes on blood flow.  One of the questions I am
invaribaly asked when giving talks about psychopharmacological imaging
studies is how we know that the effects reflect true cognitive effects of
the drugs, and do not merely represent effects on the underlying
vascalature.  If there are differential effects of the drug (relative to
placebo) on your two (or more) different cognitive states it's difficult to
continue arguing for a non-specific vascular effect.


One more point - because you can scan Ss in fMRI as many times as you like
(within reason!) it is beautifully suited to within-Ss designs, far more so
than PET with its limited number of scans.  And of course, unless you have
a cognitive reason for not scanning the same subject twice (e.g. learning,
conditioning, novelty etc.), counterbalanced, cross-over within-Ss designs
are the design of choice for psychopharmacological studies.  If you do use
a within-Ss design, remember to realign the data from the drug and placebo
sessions for each subject within a single realignment session.

My advice is GO FOR IT!

Good luck

Jenny




-----------------------------------------------------------

Dr. Jennifer T Coull
Senior Research Fellow

Wellcome Department of Cognitive Neurology
Institute of Neurology
12 Queen Square
London WC1N 3BG
UK

phone: +44 171 833 7484
fax:      +44 171 813 1420


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%