When the Target Is Also the Problem: Resolving compounding target interference in an ADA assay through heat inactivation  

SME’s: Dr. Liz Culyba, Dr. Lisette Fred Lucena, Alexandra Sullivan 

Target interference in ADA assays is one of the more persistent frustrations in bioanalysis. It distorts data in ways that are difficult to characterize, hard to mitigate, and easy to underestimate until it is too late in development. The standard toolkit addresses most cases. But not all of them. 

What follows is a case where it was not enough, and where resolving the interference required stepping outside the assay entirely to ask a different kind of question about the biology of the target itself. 

The Setup: A Target That Increases with Treatment 

Drug Y is an antibody therapeutic targeting a component of the coagulation pathway. Its target, referred to here as FVIIa, created an unusual interference scenario from the start. 

Pharmacokinetic modeling predicted that Drug Y would elevate FVIIa levels in dosed patients, with concentrations expected to reach up to 10 micrograms per milliliter. Table 1 illustrates the positive control performance in pooled matrix.  

The complication was that FVIIa was not present in the normal sample matrix at elevated concentrations, meaning it would appear at higher concentrations exclusively in dosed samples. Predose samples could not be used to assess interference. To mimic clinical samples, recombinant FVIIa was spiked into pooled matrix. The results were unambiguous. False positive signals began appearing above 0.5 micrograms per milliliter, with signal-to-noise ratios climbing steeply as FVIIa concentration increased. At the expected clinical maximum of 10 micrograms per milliliter, the signal-to-noise ratio reached 23.7 with a cut point of approximately 1.30, as illustrated in Table 2. Every sample at concentrations that would be routine in dosed patients was returning a false positive. 

The scale of the problem becomes clearer when positive control is added to the picture. At 10 micrograms per milliliter FVIIa, signal-to-noise with 100 nanograms per milliliter positive control was 27.60, compared to 23.67 without positive control. The signals were nearly additive, indicating that FVIIa alone drove samples above the cut point regardless of whether ADA was present. Any sample from a dosed patient would return a positive result, making it impossible to determine which findings reflected genuine immunogenicity and which were an artifact of elevated target concentration. 

First Attempts: Using Anti-Target Antibody 

The standard approach to target interference in ADA assays is to add an anti-target antibody that sequesters the interfering molecule before it can bridge drug or block PC binding in the assay. Because FVIIa is a well-studied coagulation factor, commercial anti-FVIIa antibodies were readily available, and multiple candidates were tested. 

None of them worked. At 5 micrograms per milliliter anti-FVIIa antibody, even with EDTA added to disrupt the calcium-dependent structure of the target, signal-to-noise ratios at 10 micrograms per milliliter FVIIa remained above 30, as shown in Table 3. The conventional approach had reached its limit. 

The Question That Changed the Approach 

Rather than continue iterating on reagent-based mitigation, the case was brought to an internal scientific office hours session, a standing forum where scientists across disciplines work through challenging assays together. The conversation shifted away from the mechanics of the ADA assay and toward something more fundamental: what was actually known about FVIIa as a protein, and whether that knowledge pointed toward a different kind of solution. 

FVIIa is a clotting factor, and clotting factors have been studied extensively in the hemophilia field, where detecting anti-clotting factor antibodies, known as inhibitors, has been a central analytical challenge for decades. One assay used routinely in that context is the Bethesda assay. A review of the relevant literature identified a useful precedent: heat treatment had been explored as a pretreatment step in commercial inhibitor assays for extended half-life clotting factor products, resulting in meaningful improvements in signal [1]. 

The hypothesis followed from there. Clotting factors, including FVIIa, may be less thermally stable than the drug antibody or the anti-drug antibody used in the assay. If samples were heat-treated before testing, it might be possible to denature the interfering target selectively while leaving the ADA signal intact.  

Heat Activation Results 

Samples spiked with FVIIa were subjected to 65 degrees Celsius for one hour before running the ADA assay. The shift in performance was substantial. 

At 10 micrograms per milliliter FVIIa, signal-to-noise in samples without a positive control dropped from 23.7 to 1.07, well below the cut point. The false positive signal was gone. Equally important, the assay retained its sensitivity to ADA. Positive control signal-to-noise, illustrated in Table 4, at the same FVIIa concentration was 5.66 after heat treatment, compared to 11.5 without heat and without interference. The reduction in signal reflected the removal of target interference, not a loss of ADA detection capacity. Positive control signal-to-noise held at 5.66, well above the cut point, confirming the assay could still reliably detect ADA.  

The target tolerance increased from approximately 0.5 micrograms per milliliter to greater than 10 micrograms per milliliter, meeting the requirement established by pharmacokinetic modeling. 

The approach worked because it targeted a specific biological property of the interfering molecule rather than attempting to block its interaction with the drug. No additional reagents were introduced into the assay matrix. Instead, heat treatment exploited the difference in thermal stability between FVIIa and the antibodies, selectively removing the source of interference while leaving detection intact. 

Problem-Solving in Bioanalysis 

The solution here did not come from running more experiments in the same direction. It came from changing the frame of the problem entirely. 

A standard ADA interference workup treats the problem as an assay chemistry issue: add something to remove the interfering substance, adjust a reagent, or modify the protocol. That is the right starting point, and it resolves in most cases. But when it does not, continuing to iterate within the same frame is unlikely to produce a different outcome. 

What changed the outcome here was bringing in contextual knowledge from outside the immediate assay: the biological nature of the target, how other fields have worked with it, and what those assays have learned along the way. That kind of cross-disciplinary thinking does not happen automatically. It requires time, relationships, and scientific infrastructure to support it. Standing forums where scientists from different backgrounds can work through difficult problems together and draw on knowledge that no single discipline holds on its own are what makes that possible. 

Heat inactivation is not a novel technique. Its application to this interference problem was not obvious within the context of the ADA assay. It became visible only when the science of the target, rather than the mechanics of the assay, became the starting point for the conversation. 

References 

[1] Payne, A. B., Ellingsen, D., Driggers, J., Bean, C. J., & Miller, C. H. (2020). Evaluation of pre-analytic heat treatment protocol used in the CDC Nijmegen-Bethesda assay for heat inactivation of extended half-life haemophilia treatment products. Haemophilia, 26(1), e28-e30. https://doi.org/10.1111/hae.13901