@DiapDealer: Does this narrow down your set enough? This one should match anything with at least one non-word(unicode) character in italics, including contractions but excluding empty spans(which should be easy enough to remove before- or afterwards.)
Code:
(*UCP)(?U)<span class="italic">[^<]*\W[^<]*</span>
If you do not want to miss absolutely anything(like nested spans) use
.* instead of
[^<]*. But you will probably match some unwanted multi-span matches.