See Tutorials: perlunitut: Unicode in Perl, perluniintro/perlunitut, and utf8, binmode, Perlmonks uses the windows-1252 (similar to Latin-1) encoding, and all characters that are not in that character set are HTML-escaped - which doesn't work inside <code>...</code> tags, because everything is interpreted literally there.

So if you put all that together

#!/usr/bin/perl --
use utf8;
use strict;
use warnings;
use XML::Twig;
use Data::Dump qw/ dd /;
my $rawl  = q{C1000 マネージャー&#
+12477;フトウェア};
my $rawe = "C1000 \x{30DE}\x{30CD}\x{30FC}\x{30B8}\x{30E3}\x{30FC}\x{3
+0BD}\x{30D5}\x{30C8}\x{30A6}\x{30A7}\x{30A2}";

for my $raw ( $rawl, $rawe ){
    my $t = XML::Twig->new( pretty_print => 'indented' )->xparse( "<a>
+<b><q></q></b></a>" );
    $t->findnodes('//q', 0 )->set_content( $raw );
    dd( $raw, $t->findnodes('//q',0)->trimmed_text, $t->sprint, );
#~     $t->print( \*STDOUT );
#~     $t->print_to_file( 'outfoo.xml' );
}
__END__
(
  "C1000 \x{30DE}\x{30CD}\x{30FC}\x{30B8}\x{30E3}\x{30FC}\x{30BD}\x{30
+D5}\x{30C8}\x{30A6}\x{30A7}\x{30A2}",
  "C1000 \x{30DE}\x{30CD}\x{30FC}\x{30B8}\x{30E3}\x{30FC}\x{30BD}\x{30
+D5}\x{30C8}\x{30A6}\x{30A7}\x{30A2}",
  "<a>\n  <b>\n    <q>C1000 \x{30DE}\x{30CD}\x{30FC}\x{30B8}\x{30E3}\x
+{30FC}\x{30BD}\x{30D5}\x{30C8}\x{30A6}\x{30A7}\x{30A2}</q>\n  </b>\n<
+/a>\n",
)
(
  "C1000 \x{30DE}\x{30CD}\x{30FC}\x{30B8}\x{30E3}\x{30FC}\x{30BD}\x{30
+D5}\x{30C8}\x{30A6}\x{30A7}\x{30A2}",
  "C1000 \x{30DE}\x{30CD}\x{30FC}\x{30B8}\x{30E3}\x{30FC}\x{30BD}\x{30
+D5}\x{30C8}\x{30A6}\x{30A7}\x{30A2}",
  "<a>\n  <b>\n    <q>C1000 \x{30DE}\x{30CD}\x{30FC}\x{30B8}\x{30E3}\x
+{30FC}\x{30BD}\x{30D5}\x{30C8}\x{30A6}\x{30A7}\x{30A2}</q>\n  </b>\n<
+/a>\n",
)
[download]