We are interested in knowing the day to day variability of HPLC assay :-

- if one cephalosporine (Cefaclor) sample analysed on ten different days using same intrument & same chemist than how much variation should be allowable?


P.R.Upadhyay
e-mail : pru60@hotmail.com

You have asked a question that has no simple answer.

In my experience, day to day variability of a HPLC assay is dependent on a number of factors. The actual instrument itself can influence variability, some give more reproducable results (RT + area) than others. The options which you may or may not have (a column heater example). The mode(s) in which you choose to run (high/low flow rates, isocratic, gradient, shallow gradients, steep gradients). Pre-mixed or instrument mixed solvents. The mobil phase/column interaction, a harsh mobil phase (pH extremes for example) that is hard on a column will infulence variability. I am sure there are many more influences that I have not thought of.

So, the answer to your question is .... you have to determine experimently what your allowable variation is with your system, columns, mobil phase, chemist, samples.

After you determine what your variability is, then your need to ask yourself some questions. Can I live with that amount of variability? If so, carry on, if not then you ask 'How can I improve my reproducibilty?' and go on from there.

A.A.

It is not clear from your message whether you are analyzing plasma/biological fluid samples or formulations.

The acceptable variation (or better precision) and accuracy for a bioanalytical method have been defined as max. ±15 % (on Quality Control samples) in the Washington consensus paper (J. Pharm. Sci. 81, 309 (1992)). See also the draft guidance on bioanalytical methods validation for human studies, available on the FDA internet site.

I do not know whether similarly defined limits have been agreed upon for the potency determination of formulations.

Addition to my previous message: for formulations, read the discussion on System Suitability on the LC board

If you told me your lab had a day to day variation of +/- 15% on your HPLC methods, I would laugh in your face and find someone else to do the work.

Dear Anonymous
probably you have never analyzed a single plasma sample in your life. Try to use your imagination to understand why the FDA thinks that +/-15% interday precision is acceptable in biological fluid analysis, or maybe read once in your life a scientific journal like Journal of Chromatography Biomedical Applications

Dear A. Guenzi

Anonymous here, and thank you for being so rude.

As a matter of fact I have analysed thousands of plasma samples, and urines, and tissues. So, I do not have to use my imagination, and, well, I have read a journal or 2 in my life. Back to the topic at hand, the original question said same sample, chemist, method. I still laugh at +/- 15% on the same sample, chemist, method. Different samples, different animals in an experiment, different studies, then +/- 15%, I can accept that. For the same samples on different day, NO WAY!

Only a fool would be inclined to step between Dr. Guenzi and anonymous. I fancy myself not a fool, and furthermore the area of discussion is not my specialty. However, is the difference in opinion here due to not defining what we are talking about? When analyzing plasma samples, a 15% variability is considered acceptable at the lowest concentration that can be quantitated. In fact, that is how this lowest concentration is generally defined. A similar convention would be appropriate for any assay where very low concentrations are involved, but where very high accuracy is not required at the lowest level. Of course, accuracy is expected to improve for higher concentrations, and in the low-concentration range accuracy (defined as one over CV) is usually proportional to concentration. Does this help?

Actually, this absolute ±15% seems very harsh if it is to apply to all analytes under all conditions. Even the German authorities, well known for their regulating prowess, allow for different ranges for various substances. In accordance, my experiences indicates that there is an extreme dependence of precision (and even more so of accuracy) on type of substance to be analyzed, and on its source. One example: Cortisol in plasma/serum of patients can be anywhere from easily analyzable to not at all, depending on the patient. In-between one gets many samples with a strong matrix (even after several, chromatographic steps) that does not chromatograph correctly, giving rise to a different pattern on each injection. Also some samples leave an excruciatingly long trail, if one does not wash the column after each injection you get interference from previous injections. After getting tired at all the suggestions on how to improve the analysis, I offered 10 000 DM to anyone that could do better on this. For over five years of keeping this bet going I had no takers. (Sorry, I no longer have this kind of money at my disposal).
Actually, instead of arguing about precision, we all should help each other to get honest accuracies.