Suppose you're writing a paper in which the three key prior results are A, B, C. Let's say that C is the most recent of the three, and discusses A and B. But C completely misrepresents the work of A and B, to the extent that it starts to undermine the very premise of C !
Now you have to discuss these prior papers: what do you do ? The conservative approach is to ignore the issue, and merely discuss A, B, and C correctly. If it starts sounding like C doesn't make any sense in the light of the correct rendering of A and B, then that's too bad.
But suppose it really bothers you that C got away with this ? Is it appropriate to mention C's misinterpretation (as politely as possible) or is it not worth it ? Would your answer be different if the paper were for a conference or for a journal ? Would the identity of the authors of C matter ? Should you just suck it up and take the high road ?