Acceptable accuracy for LC/MS/MS

<p><span>Greetings</span></p><p></p><p><span>I'm wondering what folks here consider to be acceptable limits for accuracy using LCMSMS.</span></p><p></p><p><span>I work in a clincal lab. Historically the limit has been 10%. Most procedures nowadays are performed on automated instruments using few manual steps if any at all. My lab director wants 10% for the LCMSMS as well. I have found this is not so easy. From an article I read from Validation Viewpoint the FDA recommends 15% for four of six points and 20% at the lower limit of quantitation. That seems rather wide for me but I don't see why there can't be some compromise between 10 and 15%.</span></p><p></p><p><span>As an example I am working up a Levetiracetam method on a UPLC/TQD. I get an obs error of 6.3% with concentrations 0 to 50 ug/mL. But pushing the upper limit to 75 ug/mL gets me an error up to 11%. He will not accept this. Problem is therapuetic range for this drug is up to 80 ug/mL which means we'll be diluting quite a few samples.</span></p><p></p><p><span>Any thoughts on this matter?</span></p><p></p><p><span>TIA</span></p><p></p><p><span>Mike</span></p>


  • Hi Mike,

    Is there any chromatography yet in this method? While MRM experiments usually provide sufficient specificity, I have often found that simply separating your analyte of interest from the matrix will vastly improve your RSDs and often increase the sensitivity of the method.

    - Frank

  • lizh


    I agree with what Mr. Morris stated.

    It will not be always easy to reach these 10% all the way in LC-MS/MS. FDA regulation is accepted almost everywhere... A more robust source will help to achieve these limits for a specified assay: to help to stay away from the LOQ for this assay. What MS instrument do you have?

    However, here is a possible list from one of our application chemists of things to check for the system as a rough guid.

    Detuning the Probe adjuster (spraying a little bit more far away from the Cone)

    Highering dessolvation temperature and gas flow will help to get a better dissolvation

    Eventually decreasing flow rate (especially on TQD) around 0.4 ml/min will help to get better RSD's (also linked to dissolvation)...

    Too low capillary voltage may give higher sensitivity but may increase Rsd (consider a test at 3kv and then compare with 1.5 and 1 kv, never go to 0.5kv).

    Internal standard with deuterated compounds should be used.

    Matrix effect is also something to consider, a very specific and selective sample prep will help to reduce Rsds.

    In addition to this, Full loop injection will help from the UPLC end.

    Hope this helps,


  • Thanks for the replys, I will try some of your suggestions.

    The instrument is a UPLC/TQD

    I am using deuterated internal standards but I cannot find Levetiracetam analogs and so have to use Gabapentin D-10 which comes out slightly after Levetiracetam. I am also measuring Lamotrigine and Topiramate that come out later and do not ionize as well as the early eluting Levetiracetam so I am reticent to decreasing the flow rate or capillary voltage (0.45 mL/min and 3.5 Kv).


  • Lamotrigene and Topiramate should ionize quite well with ESI. Are you using ESI Negative for Topiramate?

    Contact Schwartz Pharma for an analog of Levetiracetam. I believe they might by able to supply you with a methylated analog.

    Toronto Research Chemicals ( also has a Levetiracetam d3 in inventory.

  • We have decided not to do Topirate because of low test volume.

    Per suggestions I diluted the extract (protein precipitation) to minimize sample matrix 1:10 and injected 10 uL full loop. This prooved to be too much so I cut the injection down to 5 uL.

    I tried a lower capilary voltage of 2.5 Kv which increased sensitivity but it made precision worse so I changed it back to 3.5 Kv.

    Lastly I increased disolvation temperature to 450 (was 400) and gas flow to 900 (was 800). I don't was to mess with the flow rate (now 0.450 ml/min) as that would change my chromatography.

    CV's are less than 5.0% (was 6.0). I don't think there is much more I can do. I still get as much as 15% deviation though on my low (5 ug/mL) standard.

    I did get the director to accept FDA standard on this so in that respect the curves are acceptable.


  • This was too simple.

    I changed the "polynomial type" from linear to Quadratic curve (2nd order). Now all points nearly dead on. Even able to extend the upper range now.

    Problem solved. Wish I thought of it sooner.