Hi Will,
that is a good idea, combining regressions is always better than doing them separately (when they're not dependent on further corrections, as these shouldn't be).  You could hack the design matrix in irf_retroicor (the lines with A=[coeffs_card .... ]), note there is a flag refflag that specifies whether you're adding an extra regressor to "protect" task activation, you can just replace all this with your custom design matrix with additional vectors to regress, but also note you need to remove the later refflag stuff if you do so.  If you have any questions, or if you want me to modify the script for your situation, I'd be happy to help.  I think I'll add a script to do this in the next version as well.  Thanks Will,
Erik