Sometimes, a parser is a better tool than a regex. I like Marpa::R2:

#!/usr/bin/perl
use warnings;
use strict;
use feature qw{ say };

use Marpa::R2;

my $dsl = << '__DSL__';
lexeme default = latm => 1

:default ::=                    action => ::first
List     ::= ('<') Items ('>')
Items    ::= Item               action => list
           | Item (',') Items   action => merge
Item     ::= List
           | digits

digits ~ [\d]+

__DSL__

sub list  { [ $_[1] ] }
sub merge { [ $_[1], @{ $_[2] } ] }

my $grammar = 'Marpa::R2::Scanless::G'->new({ source => \$dsl });
my @cases=('<1>',
           '<<1>>',
           '<1,2>',
           '<1,2,<3,4>,5,6>',
           '<<<1,2>,3>,<4,5,<6>>,<<7>,8,9>>');
for my $case (@cases) {
    my $value_ref = $grammar->parse(\$case, 'main');
    use Data::Dumper; print Dumper $$value_ref;
}
[download]