Hashslices can be very efficient for sets of (non-empty) strings, so not sure why you need regexes for this...

  DB<124> @s1{a..c}=a..c
 => ("a", "b", "c")
 
  DB<125> @s2{c..e}=c..e
 => ("c", "d", "e")
 
  DB<126> %cut=%s1
 => ("c", "c", "a", "a", "b", "b")
 
  DB<127> delete @s1{ keys %s2}
 => (undef, "c", undef)
 
  DB<128> delete @cut{ keys %s1}
 => ("a", "b")
 
  DB<129> keys %cut
 => "c"
[download]

see also Using hashes for set operations...