This is, I think, another example where one may consider to discard factor 1 and to look only at factor 2 and 3.

(Of course, a better approach if possible would be baseline normalisation.)