Another option is to try to use COUPLE to modify the transition matrices, but that might take up a lot of time. At that point, it might be easier to use an external script to run a short depletion, extract out the materials you want to retain using something like OPUS (or f71tocsv), and then run another case with the same material masses that you want to keep constant along with the material masses from your depletion case. Iterate that way until you have a long enough depletion.
The problem is, like you said, you're trying to use ORIGEN in a very non-physical way, so it's going to require a good deal of work to circumvent that.