How to group pairs based on shared item in pd dataframe?

Question

I have a table below, item A,B,C,D,E are actually the same type but in the "Old_Group" column it was grouped separated by M1, M2, and M3.

Is there a way that ca