Is it possible to use back reference RegEx in Mappings?

I have a pattern I would like to remove some characters from such as this: "BuyPrice":"274 900 $",
The "Â " in the pattern should be removed to leave only "274900"

I was thinking something like FROM: ([0-9]*) ([0-9]*) TO: \1\2 TYPE: RegEx might work in the mapping section for the token.
But that does not appear to work.

Is it possible to use this sort of RegEx back reference substitution in Tokens?

Suggestions for a better way to handle this?
I was thinking perhaps a script cleanup just prior to writing out the data if the Token RegEx can not handle the transformation.
Any other ideas?

The mapping doesn't accept

The mapping doesn't accept regular expressions, but you can still use a script with "replaceAll" and that will work.