Yes, this is probably because your noise file doesn't reach zero at the end of the frequency of interest. So noise at much higher frequencies which doesn't exist in the real LDO and folds back on your signal frequency. This depends on the maximum sidebands you chose in your pnoise analysis.

Say you are interested in LDO noise until 1GHz, and lets say it is 1nV/sqrt(Hz) then the final frequency you type in the noise file is 1nV/sqrt(Hz) at 1GHz, but what you should really do is after that type 2GHz and set noise value to zero to terminate your noise file, or else your noise file will continue with 1nV/sqrt(Hz) until infinite frequencies as white noise, and folds back on you which is not the real case.