On groups with almost perfect involution