In principle, one can avoid anomaly cancellation requirement if 4th generation quarks are not like the ones from the previous three generations (for example, with vector-type couplings, not chiral)...
but it looks ugly to me, even though some people like it.
Of course, 4th generation...