It affects the posterior probabilities of the germline vs somatic classification, but these are fairly robust to errors in purity and ploidy. So if you don't see a difference, that's not unexpected.
Okay, I'm just working on some reorganization to my pipeline. I've decided to duplicate all the PureCN.R output files, one set prior to manual curation and the second after it. I think I'll run Dx.R on the manually curated output, no need to run it on the initial output.