Only data that were statistically above control levels (p < 0.05) for at least two of the doses were included.