What is your workflow that you are producing "files" from 16 bit TIFFs?  QU is meant to be used as the final stage prior to printing and if you print a 16 bit TIFF, you'll already get 16 bit gradients when you print (even when using 8 bit printer drivers).  If you are editing/converting images in QU, it should be the final pre-print phase, in which case, converted files will be 8 bits because all Windows drivers are 8 bits/channel.

The real benefit of 16 bits is when you have something like an underexposed image that needs an exposure adjustment to bring it into the 8 bit "range".  Once that's done, there is really no need for 16 bits/channel and any final edits in QU such as application of DFS or interpolation will work just as well at 8 bits/channel.

Mike