Bioanalytical Community

 View Only
  • 1.  Biological Outliers in ADA assay

    Posted 05-27-2025 11:24

    How do you go about identifying biological outliers for an ADA assay, especially for the confirmatory tier? We have a situation where there is a screening cut point of 1.9 and a high correlation between S/N and % inhibition with a confirmatory cut point of 79%. We did apply the conventional statistical outlier removal approaches but will appreciate any insight on approaches for removing biological outliers. 



    ------------------------------
    Nanda Balasubramanian
    Senior Principal Scientist
    Bristol Myers Squibb
    Cambridge MA
    [email protected]

    Disclaimer: Opinions expressed are solely my own and do not express the views or opinions of my employer.
    ------------------------------


  • 2.  RE: Biological Outliers in ADA assay

    Posted 05-28-2025 11:00
    Nanda,

    My first question would be, are these cut points so high because the average sample is high, or because the variability (SD) is high?

    Given the information you provided, I suspect one of two things - either you have a lot of true positives among your cut-point samples, or there is something causing a background of non-specific binding in your assay that can be competed by excess drug. In the past, I have taken an approach of eliminating any sample with inhibition >50% as being assumed true positive. Granted, 50% is a bit arbitrary, but this was for a population that was expected to have many positives.

    If the problem is non-specific binding, you could try adding an irrelevant antibody, or a heterophilic blocking reagent, to your assay buffer.

    John Kamerud
    JKBioanalytical Consulting LLC





  • 3.  RE: Biological Outliers in ADA assay

    Posted 05-28-2025 13:40
    Nanda,
    Both cutpoints are pretty high, though I've occasionally see this, and suggests a lot of variability in your dataset.  Of course outlier removal should reduce that variability to some extent, so...wow!  Without seeing the actual data it's hard to make cogent suggestions.  However, my definition of "biological outlier" is that is a pre-existing "true positive" and of course cutpoints should be determined on a true negative population.  The confirmatory assay contains two elements: a) screening (unspiked), b) competitive inhibition (spiked; demonstrating signal specificity); in fact one could potentially use only the confirmatory assay by itself because of this, but the tiered approach hopefully screens out most samples and thus is more efficient doing screen pos-->confirm (the latter uses twice the plate space/sample than the former).  Thus, when determining the confirmatory cutpoint I would consider any outliers to be biological outliers, ie., "pre-existing true positives", and apply that to both screen and confirm datasets.





  • 4.  RE: Biological Outliers in ADA assay

    Posted 05-30-2025 10:15

    Thanks, Eric for sharing your perspetives, I echo your response 100%



    ------------------------------
    Xiaohui Xu Ph.D.
    Director
    Daiichi Sankyo
    Basking Ridge NJ
    [email protected]

    Disclaimer: Opinions expressed are solely my own and do not express the views or opinions of my employer.
    ------------------------------



  • 5.  RE: Biological Outliers in ADA assay

    Posted 05-29-2025 11:11

    Apart from the biological outliers, the high screening CP could also indicate that the NC pool may not be appropriate for the assay and you may either want to try other pools or make a custom one using the individuals. However the high confirmatory CP is a separate issue and could be related to some specific or non-specific binding. You could explore target interference issue too, if you have any data for the circulating levels of target in population. Good Luck!  



    ------------------------------
    Alok Rathi M.S.
    Manager
    BioAgilytix Boston
    Boston MA
    [email protected]

    Disclaimer: Opinions expressed are solely my own and do not express the views or opinions of my employer.
    ------------------------------



  • 6.  RE: Biological Outliers in ADA assay

    Posted 05-30-2025 10:02

    Have you considered using the mixed model in your CP assessment to isolate high ADA population (which will be the DeFacto outlier).

    See:

    Mitigation of Pre-existing Antibodies to a Biotherapeutic in Non-clinical Species When Establishing Anti-drug Antibody Assay Cutpoint

    The AAPS Journal, Vol. 19, No. 1, January 2017



    ------------------------------
    Ritankar Majumdar
    Lead Scientist Team Lead
    LabCorp Drug Development
    Chantilly VA
    [email protected]

    Disclaimer: Opinions expressed are solely my own and do not express the views or opinions of my employer.
    ------------------------------