I've not used iTRAQ on plasma samples, however, was advised to avoid precipitating any protein sample due to inherent problems with resolubilising in an iTRAQ compatible buffer. We recently got by this problem by using centrifugal filters to buffer exchange extracted proteins in 0.5M TEAB prior to any processing. From there we performed Bradford estimations and continued with the labelling protocol as per manufacturer's advice.
Not used plasma - the analysis I performed was using crude tissue extracts, however I found the attached paper useful in terms of iTRAQ processing. The paper detailing my experiment (including the buffer exchange) has just been accepted - will forward a copy to you once I get the proofs if you think they'll help.