A PCA plot is a great way to find outliers. First and foremost, it gives you a starting point to try and figure out — why is that sample an outlier? That's the important question. Is it because there's something biologically unusual going on in the sample? Did one sample did not respond to the treatment and therefore looks similar to controls? Was the sample mislabeled? Or is it because something went wrong with the sample preparation, ex., the sample was accidentally left out overnight?
The best answer we can give you is that while a PCA plot should be used as a starting point to try and figure out what’s going on, it doesn’t definitively tell you whether to exclude that sample from the analysis.
If you do determine that it is a bad sample and an extreme outlier, you can remove the sample and rerun the differential expression analysis again without that sample. Remember, you can run as many analyses on your data as you’d like at no additional charge.