System verification

Hi everyone, 

I work in the pharmaceutical industry and I have one question about the system verification : 

Is it preferable to make a linear regression and use statiscal analysis to make sure the system is linear (even though it was verified with method validation) and use calibartion curve for caluclations
Or 
Is it enough to inject 2 standards with the same concentration and have a spec on RSD of std1 and spec on recovery between std1 & std2 ? 


What I mean is : 
If linearity was verified during validation method, is there still a possibility for the system not to be linear if you work on different HPLC for example ? 

Thanks in advance, 

Emeline