whole genome versus whole exome PON

Quick question:
Do you recommend using a PON made of whole exome samples, if the cancer dataset is whole exome,
and similarly use a PON made of whole genome samples, if the cancer dataset is whole genome?

In other words, can I mix whole exome and genome sample in my PON VCF, or is it better to maintain
two separate PON VCF, one for exome studies and other for genome studies?

Thanks!
Severine