If you have a known impurity peak that has a clear uv spectrum when ran on a PDA detector UPLC, but which gradually breaks down over a 12 month stability period, and, for a 12 months sample, the UV spectrum is strong at the apex and/or inflection points, but as a WHOLE peak, its spectrum is undefined and unfamiliar, can you reliably dismiss the peak as an unknown?

Our gradient method states that if a peak has a certain spectrum then its assigned as "X" compound, but this doesn't really account for how peaks can degrade or have other peaks co-elute over time and stressed conditions. I know co-elution can be proved by overlaying but with peaks that have mixed uv spectrum at different points of the peak, how can they be definitively termed? A purity test would obviously fail so I'm looking for opinions on how to make the call from "Its still Compound X but is breaking down" and "Because the overall spectrum is not like the method, it cant be called Compound X".