Usually not done, but it may be dependent on particular case that is type of work or your abstract, journal or conference format/rules. Adding reference is seen sometimes in abstract and conclusion part of published articles and conference papers. So, I think the exact solution may be obtain by asking, discussing with and following RR of the concern authority.
My recommendation is to follow manuscript preparation guidelines of the journal or templates than individual comments. Here, I want to share a standard manuscript preparation guidelines from IEEE, in which they clearly mentioned abstract must be 'without references'.
So, a reputed journal or publisher always has a standard manuscript preparation guidelines, otherwise you may follow that best suites for you or expert opinion, as IEEE. Thanks
Very rarely. When you really think that the citation needs to be inserted for validating some of your novel result or finding then you may but generally avoided.