I've expanded the solution so it's a bit easier to read:

$string =~ s{
                (?<![-\w./])
                (
                    \s?
                    [-\w.]*
                    $item
                    [-\w.]*
                    \s?
                )
            }{<b>$1<\/b>}gimsx;
[download]

Test output:

$ regex_slash_prob.pl
<a href="/cgi-programming-with-perl.zip"><b>cgi-programming-with-perl.
+zip</b></a>
[download]

Update:

While the solution above answers your question, consider the following.

If the HTML actually looks more like this:

my $string = qq|
        <a href="/cgi-programming-with-perl.zip">
            cgi-programming-with-perl.zip
        </a>
|;
[download]

Your output will look like:

        <a href="/cgi-programming-with-perl.zip">
           <b> cgi-programming-with-perl.zip
</b>        </a>
[download]

If you'd prefer it to look like:

        <a href="/cgi-programming-with-perl.zip">
            <b>cgi-programming-with-perl.zip</b>
        </a>
[download]

Remove both \s? lines, leaving:

$string =~ s{
                (?<![-\w./])
                (
                    [-\w.]*
                    $item
                    [-\w.]*
                )
            }{<b>$1<\/b>}gimsx;
[download]

You had them in your original so I left them in thinking they perhaps served some other purpose in the real data you're working on (as the string you posted contained no whitespace at all).