If the issue is proper motion, I think the only solution is to cross-reference the GCVS catalog with Gaia to get the full astrometric information. See an example of where this has been done for other purposes: https://academic.oup.com/mnras/article/522/4/6087/7150708Sorry I don't really know what the problem is. As said, the effect is the same in CdC, and if I choose there to use JNow instead of J2000, all is ok.
I'm not so knowledgeable in English, I can't explain it any better.