To specify code points outside of U + FFFF, you need to look for pairs of UTF-16 surrogates :
string.replace(/[\uD800-\uDBFF][\uDC00-\uDFFF]/g, '')
For future reference: one of the current ECMAScript proposals is to add the /u flag to support Additional Unicode characters that will allow:
string.replace(/[\u{10000}-\u{10ffff}]/gu, '')
source share