It would really depend on what the hacker did to the system. If it was via registry permissions or simple policy mods, you could counteract this by reapplying whatever policy you were using (assuming you maintained offline copies of it) on the DC and in the domain. If registry key permissions were used, they would be corrected when the template was applied (at least, in theory). I use imaging software to backup my servers, so I can rollback to any point and have a fully functional system. This would be ideal, but I am getting the impression that you do not have any reliable (or complete) backup system in place. In addition, if the hacker used an application on the system that had a vulnerability (such as an FTP or VPN service), you might still have a large hole to deal with anyway.
The ideal (and proper) use of DCs in a domain would include having redundant DCs (with GCs, except for the IM FSMO holder) that are *not* on the perimeter (on the Internet, hosting VPN connections, etc to reduce services that could be compromised and reduce direct contact) and behind a firewall. I am under the impression that the DC may have been acting as a webserver, which would make it much easier to attack. In theory, if you had another DC that you could use to host the GC for your domain, and then move the FSMO roles over, it could work. You would have to reapply all templates to your domain (again, assuming this was the method of modification used) before adding any new systems, and validate the integrity of all current member systems in the domain.