On the existence of a third-order phase transition beyond the Andrews critical point