I would suggest that the math doesn't hold - e.g. per the poster they tested 14 patients for G+ and there were only 10 patients triple negative.
This just means that they tested 14 patients to determine whether or not they were G+ not that there were 14 patients total that were G+, correct? I went back and looked at the poster (http://www.celldextherapeutics.com/pdf/CDX-011%20breast%20ASCO%202010%20Final%20poster%20proof%20-%20corrected.pdf ) and that's how I read it. So, that would seem to me to still leave in play the potential that a large majority of the G+ patients may have been in the triple negative group.
I am not sure about ex's math. It's tough to say. 10 out of 33 triple negative is significantly higher than the general breast cancer population at around 15% already. Thus, it is quite possible they just retested and confirmed the samples they already knew as triple negative BC.