You can open a filehandle on a piped command and you can also use Perl's built-in grep to do what sed does in your code. Here's an example applied to a couple of nonsense files.

$ diff dddd eeee
2c2
< werfef
---
> werftef
7c7
< esdd
---
> es4dd
$
$ diff dddd eeee | sed -re '/^[><]/d'
2c2
---
7c7
---
$
[download]

The script to do the same job.

use strict;
use warnings;

my $file1 = q{dddd};
my $file2 = q{eeee};

my @diffs = do{
       open my $diffFH, q{-|}, qq{diff $file1 $file2}
          or die qq{fork: diff $file1 $file2: $!\n};
       grep ! m{^[><]}, <$diffFH>;
   };

print for @diffs;
[download]

The output, which is identical to that produced by diff and sed on the comand line.

$ ./spw765446
2c2
---
7c7
---
$
[download]

I hope this is useful.