expect a lot of answers directing you towards various html scrubbers and token parsers, but here is the dirty regex:

  s{(?:<a\s[^>]+>)([^<]+)(?:</a>)}{$1}g;
[download]

there are a lot of assumptions in there, like the link label doesn't contain < (ie, <img>), there are no line breaks, HTML is well-formed, no embedded comments (<a <!--oops--> href="">...</a>), etc, etc. but if all your use cases are like the example there, it will work.