There are several causes for the presence of outlyers. Impurities, errors during the extraction, the needle that for some reason did not inject the same amount of sample, and samples that are naturally different from the other ones.
You can solve the problem using this checklist.
- check the look of the chromatograms and verify that all of them look more or less the same.
- check IS intensities and verify that their peak area is within an acceptable range for all the samples.
- normalize the data across samples.
- scale the data using UV scaling.
- evaluate the presence of outlyers using RobustPCA.
Usually doing replicates (biological, experimental and instrumental) will help you identify the cause of the outliers. In addition, will allow you to remove them without losing too much information.