Chromatography Forum

Hello.

I am in the process of validating a stability-indicating UPLC method for a pharmaceutical drug. At the moment I am in the process of evaluating calibration curves for linearity. I fitted the data to a standard least squares model using SAS JMP. The detection methods is UV (220 nm). However I am a little bit in doubt on how to interpret data. R squared is excellent, but what other parameters do I need to check to evaluate the "appropriateness" of a linear model? I would like to publish the data in a journal - but the journal does not have any guidelines. Below is the SAS JMP output.



Thank you in advance.

There is no general rule but linearity is proved in HPLC/UPLC when the 'correlation coefficient (r)' is 0.990 or more, and the y-intercept is no more than 5% of m since y = mx + b.

A spectrophotometric method can be linear if the correlation coefficient is greater than 0.950.

Hi aporesearch,

I recommend examining the distribution of the residuals in your calibration; this is a paper that may help a bit (Section 3.2.1.1.):

TrAC Trends in Analytical Chemistry, Volume 77, March 2016, Pages 167–185, "Evaluation of analytical calibration based on least-squares linear regression for instrumental techniques: A tutorial review" by Francisco Raposo.

My favorite section is the list of 29 or so explanations as to why r or R^2 values are misleading descriptors of linearity.

An alternate way to look at this is to use a plot such as recommended by Huber, here's a link to an example:

https://www.agilent.com/cs/library/prim ... 5140EN.pdf

refer to pgs. 20-21 or so.

Best Wishes! And note, JMP will handle the residuals plot(s) for you quite easily. I am still learning JMP myself, very powerful software and a good place to start working.