First of all, let us use the correct terminology and not brands. HPLC-MS/MS (or LC-MS/MS) analysis. "UPLC" is a trademark. The technique used should be described w/o trademarks as it is not brand dependent.

Your precision values sound very high overall. Without being on-site and reviewing your specific method, data and instrument configuration, no one can troubleshoot such a complex problem over the web. Precision and overall accuracy of any method are determined by a very long list of variables (too long to list here, but I will list a few). They are entirely dependent on YOUR own method, training, sample type(s), sample preparation, matrix present, instrument settings, integration settings, peak shape, the quality of the method (*Many methods I review are poor in overall quality. No HPLC method information has been provided to us by you so no constructive suggestions can be made), the condition of the instrument(s) used, column washing/flushing steps, and sample injection, to name just a few.

As with any problem analysis, you must troubleshoot using a step-by-step process checking and verifying all aspects of the method (to insure it is selective for the analytes and meets good chromatography fundamentals) and instrumentation. As you lack the training and experience needed to solve this on your own, I strongly suggest that you get the help of an experienced chromatographer in this process as a great deal of practical experience is needed to do so (and having a second or third person review it in person will save you a lot of time).

First of all, let us use the correct terminology and not brands. HPLC-MS/MS (or LC-MS/MS) analysis. "UPLC" is a trademark. The technique used should be described w/o trademarks as it is not brand dependent.

Your precision values sound very high overall. Without being on-site and reviewing your specific method, data and instrument configuration, no one can troubleshoot such a complex problem over the web. Precision and overall accuracy of any method are determined by a very long list of variables (too long to list here, but I will list a few). They are entirely dependent on YOUR own method, training, sample type(s), sample preparation, matrix present, instrument settings, integration settings, peak shape, the quality of the method (*Many methods I review are poor in overall quality. No HPLC method information has been provided to us by you so no constructive suggestions can be made), the condition of the instrument(s) used, column washing/flushing steps, and sample injection, to name just a few.

As with any problem analysis, you must troubleshoot using a step-by-step process checking and verifying all aspects of the method (to insure it is selective for the analytes and meets good chromatography fundamentals) and instrumentation. As you lack the training and experience needed to solve this on your own, I strongly suggest that you get the help of an experienced chromatographer in this process as a great deal of practical experience is needed to do so (and having a second or third person review it in person will save you a lot of time).

Mrs/Miss Li,

The reproducibility of the mass spectrometry in chemometric terms is /r/ = 1 or 100 %. Please, consider our reference [1], dealing with quantification of carbohydrates in mixture.

[1] Bioorganic Chemistry, 93 (2019) 103308

A mass spectrometric stochastic dynamic diffusion approach to selective quantitative and 3D structural analyses of native cyclodextrins by electrospray ionization and atmospheric pressure chemical ionization methods

Bojidarka Ivanova, Michael Spiteller

I would like to mention that, therein, we us our own-authored (mine and my co-author's theory according to the authorship of the works) innovative stochastic dynamic theory and model formulas, which quantify exactly the so-called fluctuations of the experimental variable 'intensity.' Please, consider more detail on our method in references [2,3], as well.

[2] Stochastic Dynamic Mass Spectrometric Approach to Quantify Reserpine in Solution; Bojidarka Ivanova, Michael Spiteller

Analytical Chemistry Letters, 10 (2020) 703-721; Received 13 Oct 2020, Accepted 16 Dec 2020, Published online: 28 Jan 2021

Download citation https://doi.org/10.1080/22297928.2020.1865834

[3] Steroids, 164 (2020) 108750

Stochastic dynamic mass spectrometric quantification of steroids in mixture — Part II; Bojidarka Ivanova, Michael Spiteller

Perhaps, the low reproducibility of the mass spectrometric results in your case is due to operator(s)'(s) random errors. There is needed an assessment of the intra- and interlaboratory variability in order to establish the factors altering the method performances.

William Letter is correct. However, as a starting point, look at what is changed each day, as I'm reading that you see a lot of variance between the same sample run on different days, while a sample run the same day shows much less variance.

William Letter already pointed out some good starting points. I just complement.

In my lab we do a full calibration in every batch if they are longer we do at least one calibration every 24 h. we typically analysed 30 compounds and use 30 isotope labelled internal standards. We usually achieve within batch SD of < 10% (multiple injections of the same sample).

The main source of error I experience is un systematic integration - but we often work close to the detection limits.

As matrix effects are systematic and usually repeatable these would not contribute to the statistical errors. However they would affect the "correctness" of the result.

There are severe issues batch to batch, especially if long operation have been done in between and dirt is piling up, especially signal intensety (hight) and thus LOQs change. However, differences in the results between the batches are not that big, but then we try hard to compensate with a lot of calibrations usuing internal standards.

So yes, depending on what is happening between Your "two meassurements" the signal hights /peak areas can change (massively) but with good IS and calibration before and after this can be kept under control

Thanks all of you William Letter Kai Bester Jack Silver Bojidarka B. Ivanova .

Following your advices, I did some trouble shooting these days. I found that technical repeatability in plasma sample is poor.

I first check the technical repeatability of targeted analytes in standard solutions and the CV% was less than 10%. However, when plasma samples were analyzed, the technical repeatability is quite poor, which confused me a lot.

The pretreatment of my plasma sample was simple. The details are as following: 500 uL MeOH containing 0.01% BHT  was added to 20 uL human plasma samples. Samples were mixed vigorously for 10S, and kept on ice for 30min for lipid extraction. Afterwards vials were centrifuged for 5min(20000×g at 4℃). Finally, 60% of the supernatants were transferred to a new vial and vacuum dried (35℃ for about 40 mins). The samples were dissolved in 60 uL MEOH for LC-MS/MS analysis.

What could be the reason?

This has been answered before. Sample prep is only part of the total method. Have your method evaluated by an experienced chromatographer, on-site, not via the web.

• "Without being on-site and reviewing your specific method, data and instrument configuration, no one can troubleshoot such a complex problem over the web."
• "As with any problem analysis, you must troubleshoot using a step-by-step process checking and verifying all aspects of the method (to insure it is selective for the analytes and meets good chromatography fundamentals) and instrumentation. I strongly suggest that you get the help of an experienced chromatographer in this process as a great deal of practical experience is needed to do so (and having a second or third person review it in person will save you a lot of time)."

Haonan Li I hope you had a nice holiday.

If you inject the same plasma sample several times, what happens? Have you done spiking experiments? If so, what are those results?

Jack Silver ,Hi.

If the same sample was injected into LC-MS/MS, the repeatability is poor(CV%>10%). I did spiking experiments using samples including spiking samples after pretreatment and before pretreatment to calculate recovery and matrix effect. The matrix effect is from 47%-303%. However, due the poor repeatability, the data is not convincing enough. And I have tried to reduce the flow rate from 0.4 to 0.1mL/min to get a better separation of targeted analytes, the response doubled. So what do you think the problem is?