Method development for the detection of anti-drug antibodies against a therapeutic peptide: assay format selection

ABSTRACT

Aim

Monitoring immune responses to therapeutic peptides with endogenous counterparts is crucial for evaluating drug safety and efficacy. In this paper, we focused on the selection of an optimal assay format to develop a sensitive, robust, and drug-tolerant immunoassay for the detection of anti-drug antibody (ADA) against a therapeutic peptide.

Results

We assessed distinct ADA assay formats for preclinical and clinical studies, such as direct binding with labeled protein A/G, direct binding with labeled multiple species-specific antibodies for detection, bridging and affinity capture elution (ACE) formats. The assay formats were evaluated based on multiple assay parameters including sensitivity, drug tolerance, individual matrix variability and inter-assay precision. Overall, direct binding assay with labeled protein A/G for detection, which utilized less labeled peptide drug and achieved desired sensitivity and drug tolerance, is appropriate for preclinical studies. Bridging assay is more suitable format to support clinical studies as bridging assay has less assay variability than ACE assay.

Conclusion

This study highlighted advantages and limitations of each ADA assay format for peptide drugs and evaluated the performance of different assay formats in the assay development process to aid in the selection of the best fit-for-purpose assay formats for preclinical and clinical phases.

Plain Language Summary

• The comparison of diverse ADA assay formats for therapeutic peptides highlights the unique characteristics of each assay format.

• The selection of a suitable ADA assay format based on the purpose and drug development phase enables the efficient development of a robust, sensitive and drug-tolerant ADA assays to support preclinical and clinical studies.

• The sensitivity, drug tolerance, matrix variability and precision are key assay parameters to evaluate an ADA assay format.

GRAPHICAL ABSTRACT

1. Introduction

Therapeutic peptides have emerged with unique biochemical properties and target specificity over recent years. Peptide drugs have been approved worldwide for the treatment of infectious diseases, cardiovascular, dysmetabolic diseases, and cancer [1,2]. Peptide drugs represent a unique category of pharmaceutical compounds and offer several advantages that bridge the gap between small molecules and biologics [3]. However, similar to therapeutic proteins, immunogenicity to therapeutic peptides can induce anti-drug antibodies (ADAs) which can potentially have an impact on pharmacokinetics (PK), pharmacodynamics (PD), efficacy, and/or safety [4–6]. Although therapeutic peptides exhibit relatively lower immunogenicity incidence compared to biologics, for the peptide drugs that have high homology to endogenous molecules, it is likely that any immune response generated against the drug will bind to the endogenous counterpart to trigger immune-mediated adverse events. ADA development is a significant concern for patients receiving peptide or protein-based therapies [7]. The recommendation from regulatory agencies is to conduct additional cross-reactivity assessments for peptide drugs that are the counterparts of endogenous molecules [4,5]. Therefore, it is crucial to develop robust, sensitive, and drug-tolerant ADA assays to evaluate the immunogenicity potential of peptide therapeutics. BI-X is a single linear peptide that consists of 8 amino acids, a half-life extension (fatty acid) and a linker. The amino acid sequence of BI-X has high homology to an endogenous peptide. BI-X has a molecular weight of 2953.5 Da. It is essential to develop suitable ADA assays to assess immune responses to support the preclinical and clinical studies.

In the clinical ADA testing, ADA assessment is typically performed using a risk-based tiered approach [8,9]. In the first tier, all samples are run in a screening assay. Samples screened positive are then analyzed in a confirmatory assay. Samples confirmed positive can be further characterized in a titration assay and a neutralization assay. Additional characterization such as cross-reactivity to endogenous counterparts may be included. In the nonclinical ADA testing, a screening analysis followed by a confirmatory analysis or a titration analysis is generally sufficient [10].

Various ADA assay formats have been explored and implemented for biologic drugs [11]. To date, the most commonly used immunoassay formats for the detection of ADA to biologics include electrochemiluminescence (ECL) based bridging assay, ECL-based direct binding assay, ACE (affinity capture elution) and solid-phase extraction with acid dissociation (SPEAD). The bridging assay is a predominant format to detect ADA for biologics. In the bridging assay, ADA binds to biotinylated and SULFO-TAG labeled drugs, forming immune complexes that are captured by a streptavidin-coated plate. Bridging assay possesses numerous advantages including low background, good assay sensitivity, simple implementation to detect ADA across different species, ability to detect most ADA immunoglobulin isotypes and high throughput testing [12,13]. However, bridging assay is susceptible to drug interference, as circulating drugs can bind to ADAs, forming immune complexes that may interfere with ADA detection, potentially leading to false-negative results. Soluble drug targets also interfere, which could potentially result in false negative or false-positive results [14]. In addition, bridging assay may not be well suited for all drug modalities. Direct binding assay uses generic detection reagents which is a considerable advantage in nonclinical studies [13,15,16]. In the direct binding assay, ADA is captured by the drug or biotinylated drug and detected using SULFO-TAG labeled multiple species-specific antibodies or SULFO-TAG labeled protein A/G. Although direct binding assay is easy to execute, it has several traits, such as a high background signal from nonspecific binding and the need of multiple species-specific immunoglobulin detection reagents, that make it less desirable for supporting clinical ADA testing [17]. Acid treatment and solid phase extraction are commonly used approaches to remove excess residual drug from the sample to improve drug tolerance [18]. ACE and SPEAD exhibit good drug tolerance which can detect ADA in the presence of a large amount of circulating drug by performing sample pre-treatment steps to extract ADA [19–21]. In the ACE assay, the ADA-free drug complexes are dissociated with acid treatment, then added to an ELISA plate coated with drug and neutralized to capture ADA on the plate. ADA is then eluted with acid and transferred to high bind MSD plate containing with neutralization buffer. The bound ADA is detected by SULFO-TAG labeled drug [20,22]. In the SPEAD assay, samples are treated with a high concentration of biotinylated drug to extract both ADA and ADA-free drug complexes. Biotinylated drug-ADA complexes are captured on a streptavidin coated plate. The ADA or ADA-free drug complexes are then dissociated from the biotinylated capture drug using acid, neutralized, and directly coated onto a high-bind MSD plate [21,22]. Compared to bridging and direct binding assays, ACE and SPEAD assays are time-consuming and complex sample pre-treatment steps potentially increase assay variability.

The development of ADA assays for therapeutic peptides presents different challenges compared to biologics due to smaller size and very hydrophobic nature of half-life extension group. It is difficult to conjugate peptide drugs with SULFO-TAG as SULFO-TAG is primarily used to conjugate protein (MW > 50 kDa) and limited free lysine residues are available in peptide drugs for conjugation. Digoxigenin (DIG) labeled peptides can be utilized as an alternative reagent to SULFO-TAG labeled peptides in bridging assay or ACE assay. However, some biotin/DIG conjugated lipidated peptides might have purity or stability issues which make them unsuitable for use in the assays. Furthermore, peptide drugs must be conjugated with carrier proteins like BSA or KLH for effective immunization to generate suitable positive controls (PC). The process of immunization for therapeutic peptides, used to generate positive controls, is significantly more time-consuming than that for biologic drugs.

In this work, we evaluated different assay formats including direct binding format with SULFO-TAG labeled recombinant protein A/G for detection (direct assay with protein A/G), direct binding format with SULFO-TAG labeled multiple species-specific antibodies for detection (direct assay with species-specific antibodies) and bridging format to detect ADA against BI-X in support of preclinical studies. Also, bridging and ACE formats were evaluated for clinical ADA testing of BI-X. Figure 1 depicts a schematic representation of each ADA assay format employed for BI-X. In bridging ADA format (Figure 1(a)), biotin-labeled BI-X peptide and DIG-labeled BI-X peptide bound to ADA to form immune complexes which were detected by SULFO-TAG labeled anti-DIG antibody. In direct binding format, the unlabeled BI-X was coated on uncoated MSD plate to capture ADA that was detected by SULFO-TAG labeled protein A/G (Figure 1(b)). Alternatively, ADA against BI-X was captured by biotinylated BI-X peptide drug and detected by SULFO-TAG labeled protein A/G (Figure 1(c)) or SULFO-TAG labeled multiple species-specific antibodies (Figure 1(d)). In ACE format with SULFO-TAG labeled anti-DIG antibody (Figure 1(e)), ADA-BI-X complexes are dissociated with acid treatment followed by neutralization and ADA was affinity captured by solid-phase biotin-labeled peptide. ADA is eluted off with acid and subsequently bound to MSD high bind plate. Bound ADA is detected by addition of DIG labeled BI-X followed by SULFO-TAG labeled anti-DIG antibody (Figure 1(f)). ADA-BI-X complexes are dissociated with acid treatment followed by neutralization and ADA was affinity captured by solid-phase biotin-labeled peptide. ADA is eluted off with acid and subsequently bound to MSD high bind plate. Bound ADA is detected by addition of biotin-labeled BI-X followed by SULFO-TAG labeled streptavidin.

Figure 1.

A schematic representation of bridging assay, direct assay with BI-X for capture, direct assay with SULFO-TAG labeled protein A/G, direct assay with SULFO-TAG labeled multiple species-specific antibodies, affinity capture elution (ACE) assay with SULFO-TAG labeled anti-DIG antibody and ACE assay with SULFO-TAG labeled streptavidin.

The objective of this study was to assess and compare ADA assay formats by evaluating key assay parameters during preclinical and clinical ADA method development stages to select the most appropriate assay format for BI-X. Presented here are results from critical optimization of assay conditions and evaluation of sensitivity, drug tolerance, individual matrix variability, and inter-assay precision of different assay formats.

2. Materials and methods

2.1. Experimental materials

Peptide drug BI-X was developed by Boehringer Ingelheim (Ridgefield, CT, USA). Affinity purified rabbit anti-BI-X polyclonal antibodies from BBI Solutions (Portland, ME, USA) were used as positive controls. The PCs were generated by immunizing rabbits with KLH conjugated BI-X and affinity purified using cysteine conjugated BI-X to purify anti-BI-X antibodies. PC 1 (lot# 864109.run2.methdev) was generated from ten rabbit sera. PC 2 (lot# 990632.2) was generated from four selected rabbit sera. The PC 1 was utilized in early preclinical ADA method development. The PC 2 was used in preclinical and clinical ADA method development.

Biotin and DIG labeled BI-X were synthesized at AnaSpec, Inc. (Fremont, CA, USA). Recombinant protein A/G was obtained from Thermo Fisher Scientific (Waltham, MA, USA). Monoclonal mouse anti-DIG antibody was obtained from Jackson ImmunoResearch (West Grove, PA, USA). Protein A/G and monoclonal mouse anti-DIG antibody were conjugated with SULFO-TAG by Boehringer Ingelheim DMPK Reagents & Mechanistic Investigation group (Ridgefield, CT, USA). SULFO-TAG anti-mouse antibody and SULFO-TAG anti-rabbit antibody, MSD GOLD 96-well Streptavidin SECTOR Plates, MSD uncoated High Bind plates, MSD GOLD SULFO-TAG NHS-Ester and MSD GOLD Read Buffer A were purchased from Meso Scale Discovery (Rockville, MD, USA).

Thirty individual treatment-naïve mouse plasma samples and forty individual treatment-naïve human serum samples were obtained from Bioreclamation IVT (Westbury, NY, USA). Negative controls (NC) were a treatment-naïve mouse plasma pool and a treatment-naïve human serum pool obtained from Bioreclamation IVT (Westbury, NY, USA). 10X PBS-0.5% Tween-20 plate wash buffer was acquired from Boston BioProducts (Milford, MA, USA). Pierce™ IgG elution buffer, pH 2.0 and Blocker^TM Casein were purchased from Thermo Fisher Scientific (Waltham, MA, USA). 100 mM glycine, pH 3.0 was purchased from Alpha Teknova, Inc. (Hollister, CA, USA). 1 M Tris-HCl, pH 8.0 was purchased from Invitrogen (Waltham, MA, USA). StrongBlock III and Next-Generation Buffer I (NGB I) were obtained from Lab Bioreagents (New Castle, DE, USA).

2.2. Anti-BI-X polyclonal antibodies generation

Ten New Zealand white rabbits were immunized subcutaneously with 500 μg of the KLH conjugated BI-X and boosted after 21 days with 250 μg of the KLH conjugated BI-X. ELISA assays were conducted to assess the binding of antibodies in rabbit sera to BI-X and measure the cross-reactivity to human IgG and mouse IgG. Bleeds from ten rabbits were pooled and purified using a Protein A column to obtain total IgG. The protein A purified antibody was further purified using cysteine-conjugated BI-X affinity column. The final product, an affinity purified antibody, was in a final buffer of 50 mM acetate with 100 mM NaCl, pH 5.0.

2.3. ECL bridging ADA assay

ADA samples were initially diluted in 100 mM glycine, pH 3.0 for 20 minutes shaking at room temperature (RT). The acidified samples were then added to a polypropylene plate followed by a neutralization buffer containing equimolar concentrations of biotin-labeled BI-X and DIG-labeled BI-X in NGB I to neutralize pH to 7.0. The plate was then incubated for 1 hour at RT with agitation. A streptavidin-coated MSD plate was blocked with StrongBlock III for 1 hour. Following the incubation, the MSD plate was washed three times using 1X PBS-0.05% Tween-20. Subsequently, 50 μL of the incubated mixture was transferred from the polypropylene plate to the MSD plate. The plate was then incubated for 1 hour with agitation at RT. After being washed three times with 1X PBS containing 0.05% Tween-20, the MSD plate was incubated with 0.15 μg/mL SULFO-TAG labeled anti-DIG antibody for 1 hour with agitation at RT. Following the incubation, the plate was washed three times using 1X PBS-0.05% Tween-20. MSD GOLD Read Buffer A containing tripropylamine was added, and the plate was read on the MSD sector imager S600. Within the instrument, a voltage was applied. In the presence of tripropylamine, SULFO-TAG participated in an ECL reaction. The electrochemiluminescent signal was directly proportional to the amount of ADA present in the matrix.

2.4. ECL direct assays with SULFO-TAG labeled recombinant protein A/G for detection

A streptavidin-coated MSD plate was blocked with Blocker^TM Casein for 1 hour at RT. Following the incubation, the plate was washed three times using 1X PBS-0.05% Tween-20 and 0.8 μg/mL biotin-labeled BI-X was added to the plate for 1 hour with agitation at RT (or an MSD high bind plate was coated with 1 μg/mL unlabeled BI-X for overnight at 4°C. After being washed three times with 1X PBS containing 0.05% Tween-20, the plate was blocked with Blocker^TM Casein for 1 hour in direct assay with unlabeled BI-X for capture). ADA samples were diluted in Blocker^TM Casein at a 1:100 ratio. After the plate was washed three times using 1X PBS-0.05% Tween-20, 50 μL diluted ADA samples were added to the plate for 1 hour shaking at RT. Following the incubation, the plate was washed three times using 1X PBS-0.05% Tween-20 and 50 ng/mL SULFO-TAG labeled protein A/G was added to the plate for 45 minutes shaking at RT. After the plate was washed three times using 1X PBS-0.05% Tween-20, MSD GOLD Read Buffer A containing tripropylamine was then added and the plate was read on the MSD sector imager S600. The electrochemiluminescent signal was directly proportional to the amount of ADA present in the matrix.

2.5. ECL direct assay with SULFO-TAG labeled multiple species-specific antibodies for detection

A streptavidin-coated MSD plate was blocked with Blocker^TM Casein for 1 hour at RT. After the incubation, the plate was washed three times using 1X PBS-0.05% Tween-20 and 1 μg/mL biotin-labeled BI-X was added to the plate for 1 hour on a shaker at RT. ADA samples were diluted in Blocker^TM Casein at a 1:100 ratio. After the plate was washed three times using 1X PBS-0.05% Tween-20, 50 μL diluted ADA samples were added to the plate for 1 hour with agitation at RT. Following the incubation, the plate was washed three times using 1X PBS-0.05% Tween-20 and 10 ng/mL SULFO-TAG anti-mouse antibody and 10 ng/mL SULFO-TAG anti-rabbit antibody were added to the plate for 45 minutes with agitation at RT. After the plate was washed three times using 1X PBS-0.05% Tween-20, MSD GOLD Read Buffer A containing tripropylamine was then added and the plate was read on the MSD sector imager S600. The electrochemiluminescent signal was directly proportional to the amount of ADA present in the matrix.

2.6. Affinity capture elution assay

2.6.1. ACE assay with SULFO-TAG labeled anti-DIG antibody

A Pierce™ streptavidin coated plate was incubated with 1 μg/mL biotin-labeled BI-X for one hour with agitation at RT. Samples was treated with Pierce™ IgG elution buffer, pH 2.0. After the streptavidin plate was washed three times using 1X PBS-0.05% Tween-20, 120 µL of acidified samples were added to the streptavidin plate and neutralized with 1 M Tris, pH 8.0 in the plate. The plate was incubated overnight at 4°C. On day 2, the plate was treated with Pierce™ IgG elution buffer, pH 2.0 for 10 minutes in order to allow for acid elution of ADA to an MSD high bind plate. ADA was neutralized with 1 M Tris, pH 8.0 in MSD plate for 3 hours shaking at RT, and the plate was blocked, washed, and incubated with 1 μg/mL DIG labeled BI-X for 1 hour shaking at RT. After being washed three times using 1X PBS-0.05% Tween-20, the plate was incubated with 1 μg/mL SULFO-TAG labeled anti-DIG antibody for 1 hour shaking at RT. The electrochemiluminescent signal was directly proportional to the amount of ADA present in the matrix.

2.6.2. ACE assay with SULFO-TAG labeled streptavidin

A Pierce™ streptavidin coated plate was incubated with 1 μg/mL biotin-labeled BI-X for one hour with agitation at RT. Samples were treated with Pierce™ IgG elution buffer, pH 2.0. After the streptavidin plate was washed three times using 1X PBS-0.05% Tween-20, 120 µL of acidified samples were added to the streptavidin plate and neutralized with 1 M Tris, pH 8.0 in the plate. The plate was incubated overnight at 4°C. On day 2, the plate was treated with Pierce™ IgG elution buffer, pH 2.0 for 10 minutes in order to allow for acid elution of ADA to an MSD high bind plate. ADA was neutralized with 1 M Tris, pH 8.0 in MSD plate for 3 hours shaking at RT, and the plate was blocked, washed, and incubated with 1 μg/mL biotin labeled BI-X for 1 hour shaking at RT. After being washed three times using 1X PBS-0.05% Tween-20, the plate was incubated with 1 μg/mL SULFO-TAG labeled streptavidin for 1 hour shaking at RT. The electrochemiluminescent signal was directly proportional to the amount of ADA present in the matrix.

3. Results

3.1. Key optimization of assay conditions in each assay format

3.1.1. Optimization of PC and conjugated reagents in bridging assay

Ten rabbits were immunized with KLH conjugated BI-X to generate anti-BI-X polyclonal antibodies. The PC 1 was produced from the ten immunized rabbit sera. PC 1 was initially used in the bridging assay for the development of ADA method in mouse plasma. However, PC 1 did not perform well in the bridging assay with low assay response of PCs. Ten individual rabbit sera were evaluated in bridging assay to select the sera which would produce anti-BI-X antibodies with high affinity to BI-X. The rabbit sera were serially diluted in assay buffer at 1:4, 1:16, 1:64, 1:256, 1:1024, 1:4096, and 1:16384 ratios. The diluted samples were screened in bridging assay. Fig. S1A showed the diluted rabbit serum 1, 2, 5, and 6 at 1:16384 ratio had higher signal-to-noise ratio than other sera. The corresponding rabbits were selected to collect more sera for purification to generate PC 2.

Two sets of biotin/DIG conjugated reagents were generated for BI-X bridging assay. The first set was produced by substituting fatty acid with either biotin-(PEG)₄ or DIG-(PEG)₄. The second set was created by incorporating a branch of biotin-(PEG)₄ or DIG-(PEG)₄ into the linker.

Both PC 1 and PC 2 were serially diluted in human serum, which was evaluated in the bridging assay under the same optimal conditions. PC 2 exhibited better assay sensitivity and higher signal-to-noise ratio than PC 1 (Fig. S1B). The two sets of conjugated reagents were assessed under the same optimal assay conditions, showing that the first set of conjugated reagents achieved better sensitivity and higher signal-to-noise ratio (Fig. S1B). PC 2 and the first set of conjugated reagents were selected for use in the evaluation of assay formats, which will aid in the development of preclinical and clinical ADA assays.

3.1.2. Optimization of direct binding assays

In direct binding formats, high background signal is generally observed due to nonspecific binding. The minimum required dilution (MRD) and the concentration of the detection reagent need to be optimized to reduce matrix effect and nonspecific binding. In the direct binding format with SULFO-TAG labeled protein A/G, the MRD was evaluated at 1:50 and 1:100 and the concentration of the detection reagent was evaluated at 20, 50, and 100 ng/mL, respectively. At an MRD of 1:100, the signal of the PC dilution curve was comparable at 50 ng/mL and 100 ng/mL of the detection reagent, as depicted in Figure 2(a). However, at a detection reagent concentration of 20 ng/mL and an MRD of 1:100, the %CV of NC was >20% (Figure 2(b)). Some individual matrices also had %CV issues. When using detection reagents at 50 ng/mL or 100 ng/mL, the NC signal response at MRD of 1:50 was higher than that at MRD of 1:100. Based on these results, an MRD of 1:100 and a detection reagent concentration of 50 ng/mL were selected and employed in the direct assay with SULFO-TAG labeled protein A/G.

Figure 2.

Evaluation of minimum required dilution (MRD) and the concentration of SULFO-TAG labeled protein A/G and evaluation of the concentration of SULFO-TAG labeled species-specific antibodies in the direct binding assays. (a) PC curve signal and (b) negative control signal were assessed under six different conditions: 20 ng/mL SULFO-TAG labeled protein A/G and MRD 1:100 (light red); 50 ng/mL SULFO-TAG labeled protein A/G and MRD 1:100 (red); 100 ng/mL SULFO-TAG labeled protein A/G and MRD 1:100 (dark red); 20 ng/mL SULFO-TAG labeled protein A/G and MRD 1:50 (light blue); 50 ng/mL SULFO-TAG labeled protein A/G and MRD 1:50 (blue); or 100 ng/mL SULFO-TAG labeled protein A/G and MRD 1:50 (dark blue). (c) PC curve signal and (d) negative control signal were assessed at three different concentrations of SULFO-TAG labeled species-specific antibodies: 10 ng/mL (black); 50 ng/mL (blue); or 100 ng/mL (gray). RLU = relative light units; PC = positive control; NC = negative control; abs = antibodies.

A high MRD of 1:100 was also employed in the direct assay with species-specific antibodies due to sufficient sensitivity and less matrix effect. The detection reagent SULFO-TAG labeled species-specific antibodies was evaluated at different concentrations of 10, 50, and 100 ng/mL. At 10 ng/mL of detection reagent, the PC dilution curve has good assay signal and good assay response ratio, which can achieve the desired sensitivity, and the background signal is lowest (Figure 2(c,d)). The detection reagent at 10 ng/mL was chosen in the direct assay with species-specific antibodies.

3.1.3. Optimization of ACE assay

In the ACE assay, we optimized the detection reagents, assay buffer, and MRD. The two different detection reagents including DIG conjugated BI-X and biotin conjugated BI-X were utilized to detect ADAs which were coated on the MSD plate. DIG conjugated BI-X and SULFO-TAG labeled anti-DIG antibody as detection reagents have lower background signal and better sensitivity compared to biotin conjugated BI-X and SULFO-TAG labeled streptavidin as detection reagents, as depicted in Figure 3(a). DIG conjugated BI-X and SULFO-TAG labeled anti-DIG antibody were selected as detection reagents for the ACE assay. Casein and 1% BSA were evaluated for the assay buffer. The use of 1% BSA as an assay buffer resulted in very high background signal, possibly due to nonspecific binding, which would make the assay unsuitable for the detection of ADAs against BI-X (data not shown). On the other hand, Casein, when used as an assay buffer, showed a low background signal and a satisfactory assay response ratio. We also evaluated MRD at 1:19 and 1:38. The signal/noise ratios at 100, 1000, 10000 ng/mL at MRD of 1:19 is higher than at MRD of 1:38 (Fig. S2). Therefore, casein and an MRD of 1:19 were the optimal conditions for the ACE assay.

Figure 3.

Evaluation of assay sensitivity in the different assay formats. (a) PC curves prepared in pooled human serum were obtained from bridging assay (green circles), ACE assay with SULFO-TAG labeled anti-DIG antibody (pink squares), and ACE assay with SULFO-TAG labeled streptavidin (black triangles). (b) PC curves prepared in pooled mouse plasma were obtained from direct assay with species-specific abs (red circles), direct assay with protein A/G (blue squares), direct assay with protein A/G for detection and unlabeled drug for capture (gray squares), and bridging assay (orange triangles). The data are presented as the PC signal obtained divided by mean negative control signal (signal/noise). PC = positive control; abs = antibodies; ACE = affinity capture elution; DIG = digoxigenin.

3.2. Comparison of ADA assay formats utilized to detect ADA against BI-X to support preclinical studies

Direct assay with species-specific antibodies, direct assay with recombinant protein A/G and bridging assay were assessed and compared by critical assay parameters including assay sensitivity, drug tolerance, individual matrix variability, and inter-assay precision using BI-X for mouse ADA studies. Assay conditions for each assay were fully optimized to ensure the comparisons made were not biased. Assays sensitivity was established by evaluating anti-BI-X antibody against BI-X in mouse plasma. An NC and eleven PC samples were analyzed in each assay. The PC concentrations were tested at 0.78, 1.56, 3.13, 6.25, 12.5, 25, 50, 100, 200, 400, and 800 ng/mL of anti-BI-X antibody. Sensitivity was determined to be the level above the assay-specific screening cut point, which was calculated based on a 5% false-positive rate. The screening cut point (SCP) and confirmatory cut point (CCP) were calculated by evaluating 30 individual mouse plasma. All three assays exhibited a good level of sensitivity below 250 ng/mL (Figure 3(b)). Direct assay with species-specific antibodies is the most sensitive, equal to 0.78 ng/mL. The sensitivities of direct assay with protein A/G and bridging assay were estimated as 1.56 ng/mL and 6.25 ng/mL, respectively. Despite the direct assays were performed at a higher sample dilution (MRD of 100) to minimize observed matrix interference, the assays are still very sensitive.

Direct assay with protein A/G for detection was further evaluated by using unlabeled drug for capture in order to eliminate the use of labeled peptide. The direct assay using peptide drug BI-X to capture ADA did not show any sensitivity at PC concentration of 800 ng/mL, as shown in Figure 3(b), because certain peptides with low molecular weight may not bind well to the surface of uncoated MSD plate. To overcome this issue, we employed a biotin-labeled peptide for capture. This peptide binds well to the streptavidin-coated MSD plate due to the strong interaction between biotin and streptavidin. This approach is likely to yield better results compared to using an unlabeled drug for capture.

Drug interference is common in preclinical toxicology studies, which are often dosed at higher drug concentrations. Bridging assay tends to be susceptible to interference by the circulating drug [19]. Appropriate preclinical ADA assays need to be developed to overcome drug interference. Three assays were tested with respect to drug tolerance. Bridging assay utilized an acidification step to separate the circulating drug-ADA immune complexes before adding labeled reagents. No acid treatment was employed in the direct assays. To assess drug tolerance, we prepared 250 ng/mL anti-BI-X PC in pooled mouse plasma in the absence or presence of BI-X at the concentrations of 25, 65, 100, and 250 μg/mL. The signal/noise ratio of 1.2 was employed as the threshold for evaluating drug tolerance. For the expected trough concentration of BI-X in the preclinical mouse samples from highest dose group, that is 65 μg/mL, the two direct assays showed a drug tolerance level ≥250 μg/mL at 250 ng/mL of PC. However, the drug tolerance of bridging assay is up to 25 μg/mL at 250 ng/mL of PC, which cannot meet the required drug tolerance 65 μg/mL for BI-X mouse studies (Figure 4(a)).

Figure 4.

Drug tolerance assessment at 250 ng/mL of PC or 100 ng/mL of PC in the different assay formats. (a) The drug tolerance samples prepared in mouse plasma were measured in direct assay with species-specific abs (red), direct assay with protein A/G (blue) and bridging assay (orange). (b) The drug tolerance samples prepared in human serum were measured in bridging assay (green) and ACE assay (pink). The data are presented as the signal obtained from each drug tolerance samples divided by mean NC signal (signal/noise). The black dash line represents signal/noise equal to 1.2. PC = positive control; NC = negative control; abs = antibodies; ACE = affinity capture elution.

Individual matrix variability refers to the variability observed in individual samples, which can be attributed to biological diversity [17,23]. The selection of assay formats can influence the variability of individual matrix and thus impact SCP and CCP. The same set of 30 individual treatment-naïve samples was evaluated in the three different ADA assays to assess individual matrix variability and determine SCPs and CCPs. Each sample was measured once for each distinct assay format. Figure 5(a,b) showed the distribution of signal/noise ratio and %inhibition of 30 individual samples in the three assays were not significantly different. No outlier was found in the bridging assay. Direct assay with species-specific antibodies and bridging assay showed minimal individual matrix variability. The individual matrix variability was slightly high, but acceptable for direct assay with protein A/G.

Figure 5.

Distribution of screening and confirmatory ADA assay signals produced by individual samples tested in the different assay formats. (A) Distribution of signal to noise ratio and (B) distribution of %inhibition values calculated by comparing assay signals produced by individual mouse samples in the direct assay with species-specific abs (red), direct assay with protein A/G (blue) and bridging assay (orange). The calculation results of SCP were 1.23, 1.33 and 1.13, respectively. The calculation results of CCP were 16.9%, 14.5% and 11.3%, respectively. (C) Distribution of signal to noise ratio and (D) distribution of %inhibition values calculated by comparing assay signals produced by individual human samples in the bridging assay (green) and ACE assay (pink). The calculation results of SCP were 1.09 and 1.12, respectively. The calculation results of CCP were 13.4% and 25.9%, respectively. SCP = screening cut point; CCP = confirmatory cut point; ACE = affinity capture elution; abs = antibodies.

The assay-specific cut points were calculated based on the analysis of 30 individual matrix samples following industry standard approaches as described in Devanarayan et al. [24]. Outliers were removed using Tukey’s box-plot method, which applies 1.5 times the inter-quartile range, through Microsoft Excel. Data normality was assessed using the Shapiro–Wilk test and skewness with JMP® version 18 software (SAS Institute Inc., Cary, NC, USA). The SCPs and CCPs for the three assays were calculated using a parametric approach, as the datasets appeared normally distributed. False-positive rates used for determining SCP and CCP were 5% and 1%, respectively, which would minimize the probability of obtaining false-negative results [25,26]. The established SCPs and CCPs for the three assays were shown in Figure 5(a,b).

Inter-assay precision of three ADA assays was evaluated using positive controls and negative controls in the method development runs. Table 1 summarized the results of inter-assay precision of the negative control, as well as inter-assay precision of positive controls. The negative control in direct assay with protein A/G has high variability compared to other ADA assays. The variability of positive controls is similar in the three ADA assays. Overall, inter-assay precision was <20% for NCs and PCs for the direct binding assays and bridging assay.

Table 1.

Summary of inter-assay precision from different ADA assay formats.

	Direct assay with protein A/G		Direct assay with species-specific antibodies		Bridging assay for preclinical studies		Bridging assay for clinical studies		ACE assay
PCs and NC	Conc. (ng/mL)	%CV	Conc. (ng/mL)	%CV	Conc. (ng/mL)	%CV	Conc. (ng/mL)	%CV	Conc. (ng/mL)	%CV
NC	0	15.8	0	4.9	0	5.2	0	3.6	0	9.2
HPC	1600	14.2	5000	9.9	10000	9.5	10000	8.8	10000	19.2
LPC	3	12.4	50	13.2	250	13.3	100	10.3	100	16.1

The inter-assay precision (%CV) of NCs and PCs was calculated from at least four method development runs which were performed on different days. Conc.= Concentration; CV = coefficient of variation; NC = Negative control; PC = Positive control; HPC = high positive control; LPC = Low positive control.

3.3. Comparison of ADA assay formats utilized to detect ADA against BI-X to support clinical studies

Due to high human matrix background signal responses observed in the direct binding format, direct assay with species-specific antibodies and direct assay with protein A/G were not chosen for clinical ADA method development of BI-X. Bridging and ACE assay formats were utilized to develop assays for detection of ADA against BI-X in human serum. Both assays were fully optimized to ensure valid assessments of sensitivity, drug tolerance, individual matrix variability, and inter-assay precision. In the bridging assay, biotin/DIG labeled reagents were evaluated at different concentrations to obtain the desired drug tolerance, using Hamilton MICROLAB STARlet liquid handler and GeNovu ADA assay optimization module. The optimization of master mix concentration indicated that the combinations of 0.1 µg/mL biotin/0.5 µg/mL DIG and 0.2 µg/mL biotin/0.2 µg/mL DIG had the best drug tolerance, which can achieve 25 μg/mL drug tolerance at 100 ng/mL of PC (Fig. S3). The signal/noise ratio at 0.2 µg/mL biotin/0.2 µg/mL DIG was higher than that at 0.1 µg/mL biotin/0.5 µg/mL DIG. Overall, the 0.2 µg/mL biotin/0.2 µg/mL DIG as a master mix combination was the optimal condition that can achieve higher drug tolerance. Additionally, an increase in the concentration of DIG labeled reagent led to a rise in the background signal, which may result in poor sensitivity.

Industry guidance recommends the development of an ADA assay with a sensitivity below 100 ng/mL for clinical studies [27]. The assay sensitivity of the bridging and ACE assays was assessed by testing serial dilutions of a positive control spiked in human serum. Relative sensitivity was 3.13 ng/mL and 50 ng/mL for bridging assay and ACE assay, respectively (Figure 3(a)). The bridging assay displayed a better signal/noise ratio of PC dilution curve and sensitivity than ACE assay. Based on PK/PD modeling of the approximate exposure with the clinical dosing, the target drug tolerance was set at 2 μg/mL for clinical samples. As previously described, if the assay does not meet the required drug tolerance, it would lead to false-negative results. The drug tolerance levels of bridging and ACE assays are 25 μg/mL and 100 μg/mL at 100 ng/mL of PC, respectively, which met the desired drug tolerance (Figure 4(b)). Although the bridging assay exhibited better sensitivity, ACE assay with the sample pre-treatment steps is able to tolerate more drug interference than the bridging assay.

Forty individual normal human samples were tested to assess individual matrix variability in bridging and ACE assays. The screening and confirmatory cut points were established by testing the same set of individual samples followed industry standard statistical approaches. The SCPs and CCPs for bridging and ACE assays were calculated based on parametric approach as the datasets in bridging and ACE assays appeared normally distributed. Distribution of the signal/noise ratio and %inhibition of individual human sera indicated that the bridging assay had very low individual matrix variability (Figure 5(c,d)). For the majority of individual samples, a signal/noise ratio ranging from 0.8 to 1.4 and % inhibition between −15% and 30% are considered acceptable. The individual matrix variability in ACE assay was higher than that in bridging assay, but it is acceptable based on the calculated SCP and CCP values of ACE assay.

Inter-assay precision of bridging and ACE assays was evaluated using positive controls and negative controls from the method development runs. The variability of positive controls in ACE assay was high, as shown in Table 1, but it still met the acceptance criteria (%CV $\le 20\mathrm{\%}$). Overall, the inter-assay precision was <20% for NCs and PCs for the bridging assay and ACE assay.

4. Discussion

Therapeutic peptides can generate ADA responses to themselves and to endogenous counterparts leading to affect PK/PD, efficacy, and safety. Assessment of ADA responses is an essential part in drug development during nonclinical and clinical evaluations [16,28]. Furthermore, selection of appropriate assay format for assay development to detect and measure ADA is important for assessing ADA responses against therapeutic peptides. Here, we presented ADA assay format selection for a therapeutic peptide based on assessments of matrix variability, assay sensitivity, drug tolerance and inter-assay precision characteristics of each assay. Each assay has been thoroughly optimized. This ensures that each assay format is evaluated under optimal assay conditions, thereby revealing its distinctive attributes.

An evaluation of individual matrix variability is crucial to assess screening and confirmatory cut points, as well as matrix interference as high variability in individual matrices could result in unsuitable assay cut points and selectivity issues. In the method development to detect anti-BI-X antibodies in mouse plasma, direct assay with protein A/G showed slightly higher matrix variability compared to bridging assay and direct assay with species-specific antibodies. This could be due to protein A/G binding with IgG, IgA, and IgM in matrix components. The estimated cut points calculated from individual samples indicated that matrix variability observed in this assay was within acceptable limits. In the method development to detect anti-BI-X antibodies in human serum, despite the less variability in individual matrix samples observed in the bridging assay, both the bridging and ACE assays indicated acceptable levels of matrix variability. Overall, the low degree of variability of individual samples was observed in bridging assay. Nonspecific binding and analytical variability could potentially lead to a slightly higher variability among individual samples in other assays.

Assay sensitivity is another key assay parameter that needs to be evaluated during ADA method development. ADA assays should be sufficiently sensitive to detect low levels of ADA. An ADA assay with great sensitivity could not only reduce interfering effects of the matrix by diluting analytical samples, but also be along with high assay drug tolerance level. In the ADA method development for BI-X, all assay formats met the sensitivity targets. A direct assay that uses either protein A/G or species-specific antibodies for detection showed superior sensitivity compared to the bridging assay, which could be attributed to the high binding affinity of protein A/G and species-specific antibodies to anti-BI-X antibody. Nonetheless, the bridging assay demonstrated better sensitivity compared to the ACE assay.

Regulatory guidelines recommend that ADA assays should be drug tolerant to expected circulating drug levels in ADA samples, otherwise circulating drug may interfere in the assays and ADA incidence would be underestimated. It is useful to select a proper assay format in terms of drug tolerance. Bridging assay with acid treatment may not achieve a high drug tolerance level. A direct assay with either protein A/G or species-specific antibodies for detection demonstrated greater drug tolerance compared to bridging assay. However, the direct binding format is not suitable for clinical ADA assays due to the potential for high levels of nonspecific binding to human matrix components caused by protein A/G or species-specific antibodies. ACE assay employed sample pre-treatment steps, which enhanced the level of drug tolerance.

Inter-assay precision is used to assess the plate-to-plate consistency, which is critical to the assessment of ADA. The slightly high degree of variability (%CV > 15%) was observed from negative controls in the direct assay with protein A/G and positive controls in the ACE assay. Especially, ACE assay included various steps where samples are transferred from plate to plate, which would potentially increase assay variability. Bridging assay and direct assay with species-specific antibodies had less assay variability.

The assessment of diverse assay formats utilized in BI-X ADA assay development, based on essential assay parameters, revealed that each format possesses its unique advantages and disadvantages, as summarized in Table 2. The choice of an appropriate ADA assay format for a therapeutic peptide should be driven by the specific requirements. These requirements will vary depending on the use of the assay and target population. Appropriate assay optimization including minimum dilution, capture and detection antibodies concentration, acid treatment, and incubation period has been undertaken to make sure a valid comparison of assay formats. The comparison of different assay formats indicated that direct assays, which employ optimal assay conditions to minimize matrix variability, bridging assay, and ACE assay could potentially serve as suitable assay formats for an ADA assay intended for a peptide drug.

Table 2.

Advantages and disadvantages of diverse ADA assay formats.

Method	Advantages	Disadvantages
ECL-Direct format (detecting with labeled protein A/G)	Better sensitivity and drug tolerance than those in bridging format High throughput Easy to apply across multiple species for nonclinical ADA assays	Potential for nonspecific binding Not suitable for clinical ADA assay due to high level of nonspecific binding
ECL-Direct format (detecting with labeled multiple species-specific antibodies)	Better sensitivity and drug tolerance than those in bridging format High throughput	Potential for nonspecific binding Isotype detection and species specificity determined by detection reagents Not suitable for clinical ADA assay due to high level of nonspecific binding
ECL-Bridging format	Detects all isotype responses High throughput Less matrix interference Easy to bridge in new lots of conjugated peptides Suitable for both clinical and nonclinical ADA assays	Need to prepare two conjugated peptides (biotin and SULFO-TAG) Vulnerable to drug and target interference
ACE format	Ability to achieve high drug tolerance Less matrix interference Suitable for both clinical and nonclinical ADA assays	Take extra labor to perform assays Time consuming less ideal for high throughput

5. Conclusions

The direct binding format offers high tolerance for drug, efficient implementation, and less dependency for labeled synthetic peptide reagents, which leads to time and cost savings. After comparing bridging format with direct binding format for BI-X in the preclinical studies, we concluded that direct binding format was well suited for preclinical ADA testing. The bridging format allowed for simple implementation to detect ADA across different species and high throughput testing. As the dose levels of BI-X in clinic are typically much lower than those in toxicology studies, the bridging format with acid treatment for clinical ADA testing may not encounter drug interference issues. The comparison of bridging and ACE formats during clinical ADA assay development suggests that bridging format that has less assay variability can be utilized for clinical ADA testing when it meets drug tolerance target. This work highlights the unique properties of each ADA assay format developed for a peptide drug, assisting in the selection of the most appropriate ADA assay formats to support its preclinical and clinical ADA assessments.

Funding Statement

All work described in this paper was funded by Boehringer Ingelheim Pharmaceuticals, Inc.

Article highlights

Background

• The development of anti-drug antibody (ADA) assays for a therapeutic peptide is challenging due to its smaller molecular size and the hydrophobic characteristics of the half-life extension group.

• The comparison of ADA assay formats is to select the most appropriate assay format for a therapeutic peptide BI-X during preclinical and clinical ADA method development stages.

Results and Discussion

• The goal of evaluating ADA assay formats, focusing on crucial parameters such as sensitivity, drug tolerance, matrix variability, and precision, is to develop a sensitive, robust, and drug-tolerant ADA assay.

• The comparison of ADA assay formats demonstrated the advantages and disadvantages of each ADA assay.

• Direct binding format, which requires less labeled reagents and has high drug tolerance, is well suited for preclinical ADA testing. Bridging format with its lower assay variability is ideally employed for clinical ADA testing.

Conclusion

• This study emphasizes the importance of selecting the most appropriate ADA assay format based on the unique properties of the peptide drug and the specific requirements of preclinical and clinical ADA assessments. Through the careful selection of an optimal assay format, we can achieve more accurate and reliable ADA testing outcomes.

Declaration of Interest

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/17576180.2025.2501937