When using PRXPARSE, how do I make it ignore some characters?

For example I'm trying to change abbreviations of gram to that full word, but I need to make SAS ignore the 100 in "100g","100gm","100gr","100 g","100 gm", or "100 gr".

I tried

retain k;
k=prxparse('s/(?#\d{1,3})(\s)?(GR\s|GM\s|G\s)/ GRAM /');
call prxchange(k,-1,test_ordered);
drop k;

It worked in some parts but I also get unwanted results like "IG GRAM" when it should remain "IGG" or "BRAG GRAM" when it should stay "BRAGG". Words ending with g turns into gram, like "BURNIN GRAM" instead of "BURNING".


When I tried:

retain k;
k=prxparse('s/ (?#\D{0,15})(?#\d{1,3})(\s)?G(R|M|\s) / GRAM /');
call prxchange(k,-1,test_ordered);
drop k;

I works when there's a blank between a number and "G" like "100 g","100 gm", or "100 gr" but nothing else.

Do you have any suggestions on how to remedy all this?