Therapeutic Product Immunogenicity Community

Hello everyone,

I would appreciate hearing your perspectives on strategies to avoid ultra-low ADA screening cut points. I have encountered discussions suggesting the use of a minimum cut point (for example, a cut point of 1.2 to incorporate the accepted CV allowance), even if that causes the FPR to fall below 5%, and leads to missing some low-titer, inconsequential ADA responses. Can anyone share their experiences/strategies on this topic?

Additionally, are there any publications/guidances on this topic that I may have missed?

Thank you

Arkadeep

Hello everyone,

I would appreciate hearing your perspectives on strategies to avoid ultra-low ADA screening cut points. I have encountered discussions suggesting the use of a minimum cut point (for example, a cut point of 1.2 to incorporate the accepted CV allowance), even if that causes the FPR to fall below 5%, and leads to missing some low-titer, inconsequential ADA responses. Can anyone share their experiences/strategies on this topic?

Additionally, are there any publications/guidances on this topic that I may have missed?

Thank you

Arkadeep

Thank you, John for your detailed explanation and reference to Robert's paper. In my past experience dealing with ultra-low cut points, I have applied several of the approaches you mentioned - like ensuring NC suitability, using S/N rather than raw signals. I have typically used 1.5*IQR for outlier removal, resorting to 3*IQR only when outliers exceed a certain %.

With these methods, I have generally observed FPR between the recommended 2-11% in the screening assay.  My concern, however, lies with the actual low titer positives that I detect due to these ultra-low cut points (which as you mentioned are not really a problem of the cut point).  In many cases, these true low-titer ADA positives lack clinical relevance. Given the recent discussions around the identification of only clinically relevant ADAs, and talks about changing the testing paradigm, my post was intended to question whether it might be time to reconsider the relevance of these ultra-low cut points.

Thank you

Arkadeep

Hello everyone,

I would appreciate hearing your perspectives on strategies to avoid ultra-low ADA screening cut points. I have encountered discussions suggesting the use of a minimum cut point (for example, a cut point of 1.2 to incorporate the accepted CV allowance), even if that causes the FPR to fall below 5%, and leads to missing some low-titer, inconsequential ADA responses. Can anyone share their experiences/strategies on this topic?

Additionally, are there any publications/guidances on this topic that I may have missed?

Thank you

Arkadeep

Hello everyone,

I would appreciate hearing your perspectives on strategies to avoid ultra-low ADA screening cut points. I have encountered discussions suggesting the use of a minimum cut point (for example, a cut point of 1.2 to incorporate the accepted CV allowance), even if that causes the FPR to fall below 5%, and leads to missing some low-titer, inconsequential ADA responses. Can anyone share their experiences/strategies on this topic?

Additionally, are there any publications/guidances on this topic that I may have missed?

Thank you

Arkadeep

Hello Arkdeep,

I'm posting this on the Immunogenicity board and the Bioanalytical board to cover both communication streams.  Apologies to those of you that are reading this twice.  John Kamerud and Robert Neely all share excellent points regarding ultra-low cut points.  You should definitely check to make sure you aren't excluding too many outliers (I start paying attention above 10% and get nervous if >15% are excluded), and making sure your negative control is representative of the cut point individuals.  

The timing of your question is excellent as I'm working on a collaborative paper with two CROs and BioData Solutions to discuss this topic in greater detail.  Our conclusion is that ultra-low cut points are not bad at all.  They don't lead to inflated rates of immunogenicity and runs don't fail more often than those with higher cut points.  Outside the scope of the manuscript's discussion is the clinical interpretation of immunogenicity and Robert Neely brings up a good point that reporting S/N can add nuanced information about the immune response superior to Positive/Negative/Titer. 

Here is a sneak peak of what is in the paper:  We tabulated cut points from 185 assays and over half of the assays had cut points at 1.20 or lower and a third of assays had cut points less than or equal to 1.10, thus ultra-low.  We performed in-depth analysis of ~12 ultra-low cut point case studies using various assay formats and drug modalities and only two required in-study cut points. In both cases there were assignable causes for the ultra-low cut points, rather than low assay and biological variability.  The first case was overzealous removal of outliers (>15%) and the second case was a negative control that was above the cut point individuals' responses.  We presented a portion of this data at EBF, so if you'd like a copy of the poster, please let me know.  

In short, ultra-low cut points were found to be quite robust and shouldn't cause undue concern for the bioanalytical scientists as long as an excessive number of outliers are not excluded, and the negative control is representative of the population.   

Robert

------------------------------
Robert Kernstock
Principal Scientist
Astellas Research Institute of America LLC
Northbrook IL
[email protected]

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

Original Message:
Sent: 10-01-2025 22:13
From: Arkadeep Sinha
Subject: Strategies to mitigate ultra-low cut points

Hello everyone,

I would appreciate hearing your perspectives on strategies to avoid ultra-low ADA screening cut points. I have encountered discussions suggesting the use of a minimum cut point (for example, a cut point of 1.2 to incorporate the accepted CV allowance), even if that causes the FPR to fall below 5%, and leads to missing some low-titer, inconsequential ADA responses. Can anyone share their experiences/strategies on this topic?

Additionally, are there any publications/guidances on this topic that I may have missed?

Thank you

Arkadeep

------------------------------
Arkadeep Sinha, PhD
Director, Bioanalytical Sciences
Upstream Bio
[email protected]

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

Thank you for reposting it Robert. With the approaches of NC suitability and controlling outlier removal, I have generally observed FPR between the recommended 2-11% in the screening assay.  As you mentioned, assay-wise, there were no issues either. My concern, however, lies with the actual low-titer positives that I detect due to these ultra-low cut points. In many cases, these true low-titer ADA positives lack clinical relevance. Given the recent discussions around the identification of only clinically relevant ADAs, and talks about changing the testing paradigm, my post was intended to question whether it might be time to reconsider the relevance of these ultra-low cut points. I would love to have a copy of your poster.

Thank you

Arkadeep

------------------------------
Arkadeep Sinha, PhD
Director, Bioanalytical Sciences
Upstream Bio
[email protected]

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

Original Message:
Sent: 10-04-2025 10:14
From: Robert Kernstock
Subject: Strategies to mitigate ultra-low cut points

Hello Arkdeep,

I'm posting this on the Immunogenicity board and the Bioanalytical board to cover both communication streams.  Apologies to those of you that are reading this twice.  John Kamerud and Robert Neely all share excellent points regarding ultra-low cut points.  You should definitely check to make sure you aren't excluding too many outliers (I start paying attention above 10% and get nervous if >15% are excluded), and making sure your negative control is representative of the cut point individuals.  

The timing of your question is excellent as I'm working on a collaborative paper with two CROs and BioData Solutions to discuss this topic in greater detail.  Our conclusion is that ultra-low cut points are not bad at all.  They don't lead to inflated rates of immunogenicity and runs don't fail more often than those with higher cut points.  Outside the scope of the manuscript's discussion is the clinical interpretation of immunogenicity and Robert Neely brings up a good point that reporting S/N can add nuanced information about the immune response superior to Positive/Negative/Titer. 

Here is a sneak peak of what is in the paper:  We tabulated cut points from 185 assays and over half of the assays had cut points at 1.20 or lower and a third of assays had cut points less than or equal to 1.10, thus ultra-low.  We performed in-depth analysis of ~12 ultra-low cut point case studies using various assay formats and drug modalities and only two required in-study cut points. In both cases there were assignable causes for the ultra-low cut points, rather than low assay and biological variability.  The first case was overzealous removal of outliers (>15%) and the second case was a negative control that was above the cut point individuals' responses.  We presented a portion of this data at EBF, so if you'd like a copy of the poster, please let me know.  

In short, ultra-low cut points were found to be quite robust and shouldn't cause undue concern for the bioanalytical scientists as long as an excessive number of outliers are not excluded, and the negative control is representative of the population.   

Robert

------------------------------
Robert Kernstock
Principal Scientist
Astellas Research Institute of America LLC
Northbrook IL
[email protected]

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

Original Message:
Sent: 10-01-2025 22:13
From: Arkadeep Sinha
Subject: Strategies to mitigate ultra-low cut points

Hello everyone,

I would appreciate hearing your perspectives on strategies to avoid ultra-low ADA screening cut points. I have encountered discussions suggesting the use of a minimum cut point (for example, a cut point of 1.2 to incorporate the accepted CV allowance), even if that causes the FPR to fall below 5%, and leads to missing some low-titer, inconsequential ADA responses. Can anyone share their experiences/strategies on this topic?

Additionally, are there any publications/guidances on this topic that I may have missed?

Thank you

Arkadeep

------------------------------
Arkadeep Sinha, PhD
Director, Bioanalytical Sciences
Upstream Bio
[email protected]

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

Hello Arkdeep,

I'm posting this on the Immunogenicity board and the Bioanalytical board to cover both communication streams.  Apologies to those of you that are reading this twice.  John Kamerud and Robert Neely all share excellent points regarding ultra-low cut points.  You should definitely check to make sure you aren't excluding too many outliers (I start paying attention above 10% and get nervous if >15% are excluded), and making sure your negative control is representative of the cut point individuals.  

The timing of your question is excellent as I'm working on a collaborative paper with two CROs and BioData Solutions to discuss this topic in greater detail.  Our conclusion is that ultra-low cut points are not bad at all.  They don't lead to inflated rates of immunogenicity and runs don't fail more often than those with higher cut points.  Outside the scope of the manuscript's discussion is the clinical interpretation of immunogenicity and Robert Neely brings up a good point that reporting S/N can add nuanced information about the immune response superior to Positive/Negative/Titer. 

Here is a sneak peak of what is in the paper:  We tabulated cut points from 185 assays and over half of the assays had cut points at 1.20 or lower and a third of assays had cut points less than or equal to 1.10, thus ultra-low.  We performed in-depth analysis of ~12 ultra-low cut point case studies using various assay formats and drug modalities and only two required in-study cut points. In both cases there were assignable causes for the ultra-low cut points, rather than low assay and biological variability.  The first case was overzealous removal of outliers (>15%) and the second case was a negative control that was above the cut point individuals' responses.  We presented a portion of this data at EBF, so if you'd like a copy of the poster, please let me know.  

In short, ultra-low cut points were found to be quite robust and shouldn't cause undue concern for the bioanalytical scientists as long as an excessive number of outliers are not excluded, and the negative control is representative of the population.   

Robert

Hello everyone,

I would appreciate hearing your perspectives on strategies to avoid ultra-low ADA screening cut points. I have encountered discussions suggesting the use of a minimum cut point (for example, a cut point of 1.2 to incorporate the accepted CV allowance), even if that causes the FPR to fall below 5%, and leads to missing some low-titer, inconsequential ADA responses. Can anyone share their experiences/strategies on this topic?

Additionally, are there any publications/guidances on this topic that I may have missed?

Thank you

Arkadeep

Hello Arkdeep,

I'm posting this on the Immunogenicity board and the Bioanalytical board to cover both communication streams.  Apologies to those of you that are reading this twice.  John Kamerud and Robert Neely all share excellent points regarding ultra-low cut points.  You should definitely check to make sure you aren't excluding too many outliers (I start paying attention above 10% and get nervous if >15% are excluded), and making sure your negative control is representative of the cut point individuals.  

The timing of your question is excellent as I'm working on a collaborative paper with two CROs and BioData Solutions to discuss this topic in greater detail.  Our conclusion is that ultra-low cut points are not bad at all.  They don't lead to inflated rates of immunogenicity and runs don't fail more often than those with higher cut points.  Outside the scope of the manuscript's discussion is the clinical interpretation of immunogenicity and Robert Neely brings up a good point that reporting S/N can add nuanced information about the immune response superior to Positive/Negative/Titer. 

Here is a sneak peak of what is in the paper:  We tabulated cut points from 185 assays and over half of the assays had cut points at 1.20 or lower and a third of assays had cut points less than or equal to 1.10, thus ultra-low.  We performed in-depth analysis of ~12 ultra-low cut point case studies using various assay formats and drug modalities and only two required in-study cut points. In both cases there were assignable causes for the ultra-low cut points, rather than low assay and biological variability.  The first case was overzealous removal of outliers (>15%) and the second case was a negative control that was above the cut point individuals' responses.  We presented a portion of this data at EBF, so if you'd like a copy of the poster, please let me know.  

In short, ultra-low cut points were found to be quite robust and shouldn't cause undue concern for the bioanalytical scientists as long as an excessive number of outliers are not excluded, and the negative control is representative of the population.   

Robert

Hello everyone,

I would appreciate hearing your perspectives on strategies to avoid ultra-low ADA screening cut points. I have encountered discussions suggesting the use of a minimum cut point (for example, a cut point of 1.2 to incorporate the accepted CV allowance), even if that causes the FPR to fall below 5%, and leads to missing some low-titer, inconsequential ADA responses. Can anyone share their experiences/strategies on this topic?

Additionally, are there any publications/guidances on this topic that I may have missed?

Thank you

Arkadeep

Arkdeep, Lauren, John, Robert N., Robert Kubiak, Joleeen, and Eric,

Once again, I'm replying on both discussion boards. I'm so thankful for your initial post and all of the responses so far.  

The publication on ultra-low cut points, as Lauren suggested, is targeted toward the Bioanalytical Scientist to give them confidence that their cut point factor is solid (reliable and assay appropriate) as long as they assess the outlier removal and NC response vs the population.  I couldn't agree more that the approach towards the assessment of unwanted immunogenicity could use a refresh.  I love the idea of looking at S/N, which in most cases follows Titer, as a means of teasing out impact of ADA on pharmacokinetics, safety, and efficacy.  Additionally, S/N has the advantage of a single sample measurement in the screening assay rather than going through the three-tiered approach cutting down on unnecessary sample analysis costs and timelines.  If I may be so bold, S/N is a surrogate for ADA concentrations, just like titer is, and who knows, in the future immunogenicity assays may have a standard curve applied to them resulting in relative ADA 'concentrations'.  To be clear, immunogenicity assays are not PK assays, they are biomarker assays, and many of the challenges with biomarker assays regarding reference standards would apply to quantitative immunogenicity assays.  I'm also not advocating for such an approach, but it is one that could be taken.  

Now to address Arkdeep's original point about clinical impact. I would say that in the case studies we explored, the rates of immunogenicity were not elevated, but I can't say for certain if those positive immunogenicity results were clinically impactful.  My guess is they were not. I'd also say that immunogenicity assessments are almost always best understood with supplemental data in retrospect.  It would be nearly impossible to say that a cut point is appropriate before conducting the clinical study (see Joleen's comment).  In my situation, and probably similarly to what is being done with Lauren at Immunologix, you start stratifying the ADA responses with PK, efficacy, and safety data.  For example, does an ADA S/N < 3 (or titers < 400) impact drug concentrations?  If not, then you've got great supporting data to suggest that while 'low-level' ADA may be detected, below a certain threshold these ADA have no apparent clinical safety/efficacy impact. Then you can state things like 20% of subjects developed ADA, but only 5% developed ADA that led to decreased efficacy.  

It is a shame that there is so much swirl when we are reporting high rates of immunogenicity where the bulk of the data shows that they are not impactful (summarizing Lauren's comment).  I've been on programs where immunogenicity results led to product termination when there were no data showing impact on efficacy or safety, nevertheless "rate of immunogenicity" was a marketing concern.  I'm so glad that Bioanalytical scientists (and AAPS) are having these discussions and publishing white papers and other examples demonstrating the use of alternative strategies to summarize the impact of ADA (e.g. S/N vs Titer, and utility of confirmatory assays).  I hope that smart science continues to lead us towards better outcomes.  

Hello Arkdeep,

I'm posting this on the Immunogenicity board and the Bioanalytical board to cover both communication streams.  Apologies to those of you that are reading this twice.  John Kamerud and Robert Neely all share excellent points regarding ultra-low cut points.  You should definitely check to make sure you aren't excluding too many outliers (I start paying attention above 10% and get nervous if >15% are excluded), and making sure your negative control is representative of the cut point individuals.  

The timing of your question is excellent as I'm working on a collaborative paper with two CROs and BioData Solutions to discuss this topic in greater detail.  Our conclusion is that ultra-low cut points are not bad at all.  They don't lead to inflated rates of immunogenicity and runs don't fail more often than those with higher cut points.  Outside the scope of the manuscript's discussion is the clinical interpretation of immunogenicity and Robert Neely brings up a good point that reporting S/N can add nuanced information about the immune response superior to Positive/Negative/Titer. 

Here is a sneak peak of what is in the paper:  We tabulated cut points from 185 assays and over half of the assays had cut points at 1.20 or lower and a third of assays had cut points less than or equal to 1.10, thus ultra-low.  We performed in-depth analysis of ~12 ultra-low cut point case studies using various assay formats and drug modalities and only two required in-study cut points. In both cases there were assignable causes for the ultra-low cut points, rather than low assay and biological variability.  The first case was overzealous removal of outliers (>15%) and the second case was a negative control that was above the cut point individuals' responses.  We presented a portion of this data at EBF, so if you'd like a copy of the poster, please let me know.  

In short, ultra-low cut points were found to be quite robust and shouldn't cause undue concern for the bioanalytical scientists as long as an excessive number of outliers are not excluded, and the negative control is representative of the population.   

Robert

Hello everyone,

I would appreciate hearing your perspectives on strategies to avoid ultra-low ADA screening cut points. I have encountered discussions suggesting the use of a minimum cut point (for example, a cut point of 1.2 to incorporate the accepted CV allowance), even if that causes the FPR to fall below 5%, and leads to missing some low-titer, inconsequential ADA responses. Can anyone share their experiences/strategies on this topic?

Additionally, are there any publications/guidances on this topic that I may have missed?

Thank you

Arkadeep