use v5.14;

my $string = 'Lorem ipsum dolor sit amet, consectetur adipisicing elit
+.';
my @skips  = ( 'ipsum', 'sit amet', 'elit' );

my $re = do {
  local $" = '|';
  qr/@skips/;
};

say for $string =~ s/$re//gr =~ m/\b(\w+)\b/g;
[download]

Output:

Lorem
dolor
consectetur
adipisicing
[download]

Update (explanation): The s///r construct returns a new string lacking all "skip" words/phrases. The new string is fed immediately into the second regexp that matches complete words (at least as defined by "\w"). A list is formed, and those are output using 'say' in a loop. If you want to store them instead, change that last line to:

my @keeps = $string =~ s/$re//gr =~ m/\b(\w+)\b/g;
[download]

...or...

my $VAR = [ $string =~ s/$re//gr =~ m/\b(\w+)\b/g ];
[download]

...if you prefer holding a reference to an anonymous array.