split removes the separator from the resulting tokens:

      my $str = "1\n2\n3\n4";
      my @tokens = split( /\n/, $str );
      # now tokens contain
      # ( "1", "2", "3", "4" ), NOT ( "1\n", "2\n", "3\n", "4\n" )
[download]

And no, $\ is not the output record separator, the output record separator is $,

   Update Above was not true, $\ is output record separator.
   $, was output field separator

But then again, with your code I don't understand the reasoning to use split. you could just

    my $para = ..... # whatever.
    # so $para is ( possibly ) a multi-line parameter

    my $first_line_break = index( $para, "\n" );
    if( $first_line_break > 0 ) {
         $lastHeading = substr( $para, 0, $first_line_break );
         next;
    }

    if( $para =~ /mSOriginating/ && $para =~ /$msisdn/ ) {
           print $para;
     }
[download]

This would do, no?