Drug‐Tolerant, Chemiluminescent Lateral Flow Immunoassay Platform for the Determination of Neutralizing Anti‐Drug Antibodies

Abstract

Administration of mAb therapeutics often elicits the production of neutralizing anti‐drug antibodies (nADA) in patients. Current nADA assays are drug‐sensitive and detect only free nADA, thereby overlooking drug‐bound antibodies and reducing sensitivity. An extra preparatory step of acid dissociation is often applied to resolve the immunocomplexes and subsequently detect the resultant free nADA separately. This tedious sample preparation limits them to lab settings and exposes the antibodies to harsh conditions. Here, a modified lateral flow assay platform, namely sequential binding flow immunoassay (SBFIA) is presented for nADA determination that circumvents the drug's interference through a distinctive approach. Drawing inspiration from chromatography, sequential binding events (competitive and direct) are exploited to eliminate drugs’ interference, enabling sensitive detection of drug‐bound nADA. The platform is optimized using Infliximab (IFX) as a drug and neutralizing anti‐Infliximab antibodies (nαIFX) as a model analyte. After optimization, a minimum concentration of 103 ng mL⁻¹ nαIFX is detected. The assay shows remarkable drug tolerance and specificity to nαIFX. Based on its point‐of‐care (POC) advantages while avoiding harsh, acidic sample pretreatment, the assay can assist physicians in rapid and accurate decision‐making. By tweaking the bioreceptors, the configuration can adapt well to other biologics and respective nADA.

Immunogenicity evaluation in patients receiving mAb therapy is becoming the standard of care in healthcare systems worldwide. Nevertheless, current methods for anti‐drug antibody determination suffer from drug interference and/or inability to be performed at the point‐of‐care (POC). Herein, an engineered new platform, namely, sequential binding flow immunoassay, is conceptualized, which utilizes a unique configuration to circumvent drug interference, allowing the quantification of anti‐drug antibodies with high sensitivity, at the POC.

1. Introduction

Therapeutic monoclonal antibodies (mAb) represent a major class of biological drugs and have been extensively administered to treat conditions such as snakebite envenoming, infections, cancer, and autoimmune diseases. However, the long‐term repeated administration of mAb in patients induces immunogenicity, consequently leading to a loss of response (LOR) due to hastened clearance and/or blocking of drug‐binding sites mediated by anti‐drug antibodies (ADA).^{[ 1} , ^{2 ]}

Infliximab (IFX) is a chimeric mAb used effectively as a tumor necrosis factor inhibitor (TNFi). ≈600 000 patients receive or are between IFX infusions on any given day.^{[ 3 ]} Although the incidence of ADA development varies across clinical studies, recent data suggest that up to 83% of patients undergoing continuous IFX therapy may develop ADA.^{[ 4} , ^{5 ]} ADA against IFX (αIFX) can be stratified into neutralizing antibodies (nαIFX) and binding antibodies (bαIFX). nαIFX blocks the complementarity‐determining region (CDR) of IFX and accordingly interferes with the drug's activity as well as accelerates its clearance.^{[ 6 ]} Meanwhile, bαIFX targets epitopes outside the drug's CDR site, thus only reducing the drug's bioavailability through its clearance.^{[ 7 ]} Given their effect on drug pharmacokinetics, ADA detection is paramount for optimizing IFX therapy. Clinical studies have demonstrated that the presence of αIFX is associated with markedly lower trough IFX levels and a significantly higher risk of LOR, with meta‐analyses indicating a ≈3.2‐fold increased risk in patients with inflammatory bowel disease (IBD).^{[ 8 ]} Notably, assays that specifically quantify nαIFX have shown superior predictive power for mAb treatment failure compared to assays measuring total αIFX.^{[ 9 ]} Thus, quantifying nαIFX provides mechanistic insight into patients who have lost response, facilitates dose optimization or treatment switching, and enables more personalized treatment strategies. The immunogenicity models suggest that IFX‐nαIFX immune complexes dominate the early phase of ADA response, rather than free circulating nαIFX.^{[ 10 ]} Further, assays that “count” the former have proven to reveal the onset earlier.^{[ 11} , ^{12 ]} Recently, it has been reported that a timely uncovering of immunogenicity against IFX allows earlier treatment decisions and correlates with improved treatment outcomes.^{[ 4} , ^{13 ]}

Previously, different assay formats have been developed for ADA detection, such as Radioimmunoassay,^{[ 14 ]} various formats of enzyme‐linked immunosorbent assay (ELISAs),^{[ 15 ]} Cell‐based assay,^{[ 16 ]} Surface plasmon resonance (SPR),^{[ 17 ]} and a qualitative Lateral flow assay (LFA).^{[ 18 ]} However, except for the LFA, most platforms require centralized laboratory infrastructure, technical expertise, extensive sample preparation, and long turnaround times. Moreover, they are prone to drug sensitivity, meaning their accuracy is undermined by circulating IFX in patient serum. To address this, pre‐treatment steps such as acid dissociation have been incorporated into ELISA^{[ 19 ]} and electrochemiluminescence‐based assays (e.g., Precipitation and Acid Dissociation, or PandA).^{[ 20 ]} These approaches have successfully revealed ADA responses earlier than conventional drug‐sensitive formats.^{[ 11} , ^{12 ]} Nevertheless, this additional dissociation step prolongs the already tedious procedures and subjects the sample to harsh acidic conditions, raising concerns about whether it is suitable for clinical practice.^{[ 10 ]}

Lateral flow immunoassay (LFIA) technology has thrived over the past decades and aligns well with the World Health Organization's ASSURED criteria (affordable, sensitive, specific, user‐friendly, rapid and robust, equipment‐free, and deliverable to end users).^{[ 21} , ²² , ^{23 ]} LFIAs are now used across many sectors, including biomedical, environmental, food, and agriculture. One application of LFIA consists of antibody determination. Typically, these LFIAs are developed to screen infected or recovered individuals for Ig (mostly IgG & IgM), employing a classic sandwich configuration wherein anti‐isotype antibodies (e.g., αIgG or αIgM) are immobilized at the test line to capture antibody‐labeled conjugates from the sample.^{[ 24 ]} Each line captures the specific isotype previously bound to a labeled bioreceptor on the conjugate pad (CP), creating a sandwich. During the SARS‐CoV‐2 pandemic, LFA technologies were developed to determine neutralizing antibodies,^{[ 25} , ²⁶ , ^{27 ]} though these assays did not face nor address the issue of drug interference. To detect nαIFX, researchers are faced with the following challenges that preclude them from utilizing the classic LFA configurations: (I) nαIFX can exist in multiple isotypes and subclasses; (II) IFX, as a chimeric mAb, partially shares human immunoglobulin sequences, thus resembling host Immunoglobulins and complicating selective detection; (III) serum nαIFX exists both in free form and bound to IFX; and (IV) serum contains IFX along with nαIFX. Although significant developments have been made toward labeling molecules & signal enhancement,^{[ 28 ]} biorecognition elements,^{[ 29 ]} multiplexing,^{[ 30 ]} and quantification,^{[ 31 ]} the principles of “competitive” or “direct” binding remain at the core of most LFA. As such, the classic LFA setup limits its application to specifically designed assays; therefore, the complex case of detecting nαIFX in the presence of IFX remains a challenge.

Herein, we introduce a novel lateral flow configuration, namely, sequential binding flow immunoassay (SBFIA), designed for the rapid and drug‐tolerant detection of nαIFX. This platform integrates sequential binding events under capillary‐driven flow to enable the detection of the analyte of interest. The core principle is based on first saturating the free nαIFX of the sample with IFX on the sample pad (SP) and, further, utilizing a multi‐layered protein “hurdle concept” in the analytical membrane to eliminate the interferent biomolecules (IFX and marker antibody), allowing the spatial separation of the nαIFX‐IFX. The platform was meticulously optimized for nαIFX quantification and benchmarked against ELISA. Additionally, drug tolerance and specificity tests were carried out. The testing process was completed in less than 30 min, using widely available readers. To the best of our knowledge, this is the first rapid, drug‐tolerant assay capable of specifically and sensitively quantifying nαIFX. Thus, we envision that this platform has great potential for clinical deployment to aid physicians in better understanding the immunogenicity status of patients and improving therapy outcomes. Due to its generic nature, the platform can be readily adapted to other biologics and respective ADAs.

2. Results and Discussion

2.1. Principle of the Immunoassay

The schematic illustration of SBFIA for nαIFX detection is presented in Figure 1 . At the proximal end of the strip, an SP impregnated with IFX allows saturation of the incoming free nαIFX. Subsequently, the CP impregnated with Reporter (R) ensures the marking of biomolecules. Next, the stack pad (STP) promotes enhanced mixing of the immunoreagents. The nitrocellulose membrane (NC) features three distinct interaction protein areas, ensuring the stratification of the resulting biomolecule complexes. The first area consists of a single TNF line, the second consists of three IFX lines, and the third consists of a single Anti‐Human Fc (AHFc) line.

Figure 1.

Schematic illustration of the immunoassay's configuration during a run test with a sample containing IFX and nαIFX. Image created with BioRender.

Figure 1 shows the loading of a sample containing IFX, nαIFX, and bαIFX in the SP during the run. Both nαIFX and bαIFX bind to the bedding IFX, creating nαIFX‐IFX and bαIFX‐IFX immunocomplexes. Next, the running buffer (RB) is applied to initiate immunoreagent flow from the SP to the CP and STP matrix. As the immunoreagents flow from the SP to the CP matrix, the free IFX, nαIFX‐IFX, and bαIFX‐IFX immunocomplexes bind to the R. The complex formation between the molecules of nαIFX, IFX, and R is guided by the association reaction (K _on ) of nαIFX to IFX (with an affinity K_D = 0.12 nм), and R to IFX (K_D = 0.7 nм), to reach the equilibrium. The association between the three molecules can be described as dynamic upon liquid application and the start of capillary flow in the porous pads. For a given period, higher target concentrations yield more complex formations. Within the assay's working concentration range, nαIFX binding reaches saturation, ensuring reliable complex formation. Likewise, in classic LFIAs, the reaction is predominantly driven by mass transport, and free, unlabeled analyte molecules do not generate or interfere with the signal during the measurement. Figure S7 (Supporting Information), confirms that nαIFX and R bind to distinct IFX epitopes. R associates with IFX by targeting an epitope out of the CDR region, conversely to nαIFX, which binds to its CDR region, preventing TNF binding.

Further, the resulting complexes (IFX‐R, nαIFX‐IFX‐R, bαIFX‐IFX‐R, and unbound R) migrate to the NC, wherein they interact with the protein areas. The complex formation in the NC membrane between the incoming molecules and the capturing agents is guided by the association reaction of the partners, similarly to the one described for the pads. The first interaction is with the TNF line, which captures IFX‐R and bαIFX‐IFX‐R; the second is with the IFX area, which captures unbound R; and the third is with the AHFc line, which captures nαIFX‐IFX‐R. Eventually, by adding the substrate solution, an entirely Horseradish peroxidase (HRP)‐dependent signal is generated, whose strength correlates with the concentration of the analyte. This technology proves that nαIFX can be discerned from similar biomolecules that share the same host by creating a sequence of binding events combined with the capillary flow to separate them in an affinity chromatography‐like strip setup.

2.2. Optimization of the Assay

2.2.1. Optimization of the Capture‐Filtration Area

Establishing an area that would capture and filter the excess Reporter was crucial; therefore, different parameters of the immobilized IFX area were optimized. As shown in Figure 2a, a higher surface area of immobilized IFX enabled a lower background signal at the AHFc line. This signal reduction was proportional to the protein surface area, with the four IFX lines accounting for the lowest background signal (≈1400 relative light units ‐ RLU). Accordingly, the highest number of IFX lines was employed in the subsequent experiment (Figure 2b), and the effect of IFX concentration on Reporter capture‐filtration was tested.

Figure 2.

Reduction of background signal in the AHFc line by optimization of IFX capture‐filtration protein lines parameters. a) Effect of the number of IFX capture lines on AHFc line strength. b) Effect of IFX capture lines’ concentration on AHFc. Bars and error bars represent mean ± SD of three replicates (n = 3). RL‐Relative light. **P‐value <0.005, Kruskal‐Wallis comparison test.

Furthermore, as the Reporter migrated through the analytical membranes, a gradual decrease in signal strength was observed (Figure 2a; Figure S2a, Supporting Information), suggesting that a hurdle concept could be employed successfully to decrease false positives by reducing the amount of traversing Reporter that would otherwise migrate to the endline. The concentration of immobilized IFX also affects the signal‐to‐noise ratio at the AHFc. As shown in Figure 2b, higher concentrations of IFX in capture‐filtration lines were associated with a low background signal at the AHFc, and the noise decrease followed a dose‐response. These findings are consistent with our previous report, corroborating our earlier findings that functionalized pads can effectively block unwanted immunoreagents.^{[ 32 ]} Noteworthy, in that report, several nitrocellulose pads functionalized with bacteria were utilized to prevent false positives in the reaction zone of an E.coli vertical flow detection system. These results suggest that tweaking protein lines on a membrane could be exploited to block the migration of undesired immunoreagents in complex assays.

To minimize the amount of IFX reagent per strip, a setup comprising four IFX lines, each 2 mg mL⁻¹, was compared to a setup with three IFX lines, each 2.4 mg mL⁻¹. A slight but statistically insignificant difference was observed; thus, the format with three lines of 2.4 mg mL⁻¹ was selected as the optimal condition (see Figure S2b, Supporting Information for the noise reduction by combining IFX area parameters).

2.3. Sensitivity of the System in Detecting nαIFX

To test SBFIA's performance in nαIFX detection, strips were prepared in triplicate and run with different concentrations of nαIFX‐spiked buffer (nine‐point dilution, ranging from 51 to 13 300 ng mL⁻¹). Two negative controls were included: C1 comprised a strip setup with no inherent IFX and ran with a sample containing nαIFX at 6650 ng mL⁻¹; C2 comprised a strip setup with inherent IFX as the other positive setups and ran without nαIFX in the sample. Overall, the treatments show increased signal strength on the AHFc line in samples containing nαIFX compared to samples without nαIFX (Figure 3 ). This marked a milestone as the test could confidently discern positive and negative samples.

Figure 3.

Detection of nαIFX concentrations (from 51 to 13 300 ng mL⁻¹) in spiked samples. Bars, data points, and error bars represent mean ± SD of three replicates (n = 3). a) Dose response of the AHFc line (inset: an image of single strips side by side for tested concentrations); C1 ‐ strips with no inherent IFX and run with a sample containing nαIFX at 6650 ng mL⁻¹; C2 ‐ strips with inherent IFX and run without nαIFX. b) The fitted curve corresponds to the following equation: y = A2 + (A1‐A2)/(1 + (x/x0)^p) with adjusted R² = 0.94 (inset: linear detection range R² = 0.98).

Furthermore, a linear fit was achieved in AHFc signal intensity for nαIFX detection in the range between 200 and 6600 ng mL⁻¹ with R² = 0.98 (Figure 3b). The platform's dynamic range extended from 103 to 6650 ng mL⁻¹, the point at which a plateau was reached, and the signal was saturated. The signal saturation is likely caused by the appearance of the hook effect, a well‐known phenomenon in immunoassays,^{[ 33} , ^{34 ]} which limits the dynamic range at the upper level. Previous studies regarding αIFX presence in patients focus mainly on the prevalence of positive versus negative cases, based on a cut‐off value for αIFX, rather than reporting the absolute values in each individual.^{[ 35 ]} However, a few reports indicate that ≈90% of the positive cases for αIFX range from 0.1 to 10 µg mL⁻¹.^{[ 17} , ³⁶ , ^{37 ]} Additionally, in clinical practice, αIFX (8 µg mL⁻¹) has been used as a cut‐off value for clinical decision making.^{[ 36 ]} Following the abovementioned figures, a sample dilution (e.g., tenfold) would be required for precise quantification at SBFIA's dynamic range. Nevertheless, other studies have defined αIFX (3 µg mL⁻¹) as sufficient for clinical decision making.^{[ 38 ]} The high sensitivity and dynamic range of SBFIA allow sample dilution, thus projecting its adoption in algorithms with differential cut‐off values.

Figure 3 shows the capability of the system to measure nαIFX quantitatively. As demonstrated from Figure 3a inset, an increase in the nαIFX concentration caused a reduction of signal strength in the TNF capture line, indicating an inhibition of the inherent bedding IFX. Additionally, it reveals that the nαIFX‐IFX‐R immunocomplexes avoided the TNF line and further migrated toward the AHFc line. In principle, all the foreseen interactions in the pads were successful, and a dose‐response was achieved, thus proving the assay concept. Altogether, these results suggest that the developed platform could successfully detect nαIFX. Moreover, it exhibits a higher threshold sensitivity by at least an order of magnitude compared to the cut‐off point for decision‐making in clinical settings.^{[ 36} , ^{38 ]} Theoretically, the sensitivity could be further improved using monovalent antibodies or by minimizing the background signal using higher concentrations of immobilized proteins. Nevertheless, the present assay shows a higher sensitivity by order of magnitude than the only existing qualitative, drug‐sensitive gold nanoparticle (AuNP) LFA for ADA detection.^{[ 18 ]} A detailed comparison table, including other non‐POC methods, is provided in Table S1 (Supporting Information). In patients receiving mAb, nαIFX are produced due to an immunological reaction against the IFX, and their concentrations in sera increase over time if the same treatment is applied. As such, a sensitive, drug‐tolerant assay for nαIFX is clinically highly relevant as it can uncover the loss of response earlier and provide physicians with meaningful data to change the treatment regimen.^{[ 35 ]} Thus, SBFIA can potentially fill this gap due to its demonstrated sensitivity.

2.4. Drug‐Tolerance of the Platform Assay

The present platform is based on the saturation of incoming nαIFX with the inherent pad‐embedded drug. To test the setup's applicability in IFX‐containing samples, the RB was spiked with 2.4 µg mL⁻¹ nαIFX and, simultaneously, with concentrations of IFX varying from 625 to 20 000 ng mL⁻¹. Strips were prepared as described in the standard procedure. A negative control comprised a strip containing inherent IFX but no additional IFX or nαIFX from the sample. A positive control was set up with inherent IFX and without IFX but with added nαIFX. In all treatments but the negative control, nαIFX concentration was 2.4 µg mL⁻¹.

Figure 4 shows that the signal measured at the AHFc line in all the positive samples is higher than that of the negative control. Although a slight difference is observed in the signal intensity among positive treatments, this is not statistically significant. This reduction is likely caused by the competition of excess IFX with complexes of IFX‐nαIFX for binding the Reporter, resulting in fewer marked complexes per AHFc line in the samples containing excess IFX. This is also confirmed by the low signal at the IFX capture‐filtration lines, where the signal strength reduces as the concentration of the sample added IFX increases. Furthermore, the signal intensity in the TNF line is absent at lower concentrations of excess IFX (625 to 2500) and apparent at higher concentrations (5000 to 20 000 ng mL⁻¹). It is therefore confirmed that in the high IFX concentrations, the drug is active and non‐neutralized by the tested concentration of nαIFX. Interestingly, the platform shows remarkable drug‐tolerance in the subtherapeutic (< 3 µg mL⁻¹), therapeutic (3–7 µg mL⁻¹), and supratherapeutic ranges (> 7 µg mL⁻¹), with the highest tested concentration being 20 µg mL⁻¹ IFX. Altogether, these results suggest that the proposed platform is not affected by the presence of the drug in the sample and has excellent potential to be applied as a drug‐tolerant assay. Several assays have been reported as drug‐tolerant, such as PandA based on Meso scale discovery (MSD)^{[ 20 ]} and ELISA,^{[ 19 ]}; however, they require tedious sample preparation procedures or nonstandard benchmark equipment and expose the nαIFX to harsh acidic conditions to dissociate immunocomplexes. Others, like the cell‐based assay,^{[ 39 ]} tolerate only low amounts of IFX (<650 ng mL⁻¹). As such, bridging drug‐sensitive ELISA has been the gold standard procedure widely applied in the clinics.^{[ 35 ]} SBFIA utilizes a unique configuration that eliminates drug interference during nαIFX detection, thus showing potential for clinical application.

Figure 4.

Determination of the platform's drug‐tolerance capability. a) Image of single strips side by side for tested concentrations. b) Graphical representation of signal intensity measured in the AHFc line. Bars and error bars represent mean ± SD of three replicates (n = 3). ns: P‐value = 0.076, Kruskal‐Wallis with Dunn's multiple comparison test.

2.5. Specificity Testing of the Optimized Setup

nαIFX antibodies have been reported to significantly reduce the drug's bioavailability in patients’ circulation through clearance and inhibition. Conversely, bαIFX antibodies only enhance clearance. During the latter binding, the drug can still be active and cleared after cytokine binding. Hence, to discern between the two types of αIFX antibodies, SBFIA was designed utilizing a competition binding step. To prove the concept, we tested three different bαIFX antibodies, bαIFX_a1, bαIFX_a2, bαIFX_a3, and a combination of three bαIFX_a1+a2+a3 (see Experimental section, Specificity of the optimized setup). A negative control was set up with inherent IFX and without nαIFX or bαIFX in the sample. A positive control comprised the same strips as previously described; however, it ran with nαIFX in the sample and without bαIFX. The concentration of each αIFX antibody in all the above treatments was 2.2 µg mL⁻¹.

Figure 5 shows that the sample containing nαIFX scored a higher signal intensity in the AHFc line than all bαIFX. Conversely, the bαIFX tests show higher signal intensity on the TNF capture line, indicating that the inherent IFX was not neutralized; thus, the complexes of R‐IFX‐bαIFX were successfully captured on this line. These results demonstrate that SBFIA is effective in discerning between nαIFX and bαIFX antibodies. This feature of the platform can be utilized in future POC tests to screen for immunization efficiency, a significant parameter in public health. The SBFIA strips were initially stored at room temperature (relative humidity ≈40%) and tested after one week. A total signal loss was observed, likely attributed to the Reporter's HRP activity loss at room temperature. This is in accordance with the manufacturer's instructions on the Reporter's usage.^{[ 40 ]} Conversely, when SBFIA test strips were stored in desiccated, sealed lightproof pouches at −20 °C, no detectable signal loss was observed for over 30 days. Nevertheless, we recommend that stability be determined empirically on a case‐by‐case basis in future platform applications, due to the differences in HRP isozymes and conjugation chemistries used in labeling antibodies.^{[ 41} , ^{42 ]}

Figure 5.

Determination of the platform's specificity. a) Image of single strips side by side for the different αIFX antibodies. b) Graphical representation of signal intensity measured in the AHFc line. Bars and error bars represent mean ± SD of three replicates (n = 3). ns: P‐value = 0.16, Kruskal‐Wallis with Dunn's multiple comparison test. ^*P‐value = 0.032, Kruskal‐Wallis comparison test.

2.6. Human Sample Analysis

The levels of serum immunoglobulins (IgA, IgM, IgG) vary among healthy adults, with typical concentrations between 7‐19 mg mL⁻¹. Although the irrelevant Ig in sera could be captured in the AHFc line, the assay was conceptualized in such a way that these Ig would not cause an increase in signal intensity due to not being marked by the Reporter during the sequential interactions. Figure S6a (Supporting Information) shows that sera samples spiked with nαIFX had lower AHFc intensity than the RB counterparts, indicating that the low intensity is caused by the irrelevant Ig occupying the binding sites of AHFc. As expected, the irrelevant sera Ig competed with the complexes of nαIFX‐IFX‐R for the AHFc line but did not cause an increase in signal intensity or false positives. Although it contributed to weakening the signal intensity, the ratio between positive and negative samples was maintained (see Figure S6a, Supporting Information for the serum effect on signal intensity). To optimize the setup for sera screening, the issue was circumvented by increasing the AHFc concentration to 1.8 mg mL⁻¹ on the immobilized line (Figure S6b, Supporting Information).

Compared with a validated Anti‐Human λ chain ELISA format^{[ 37} , ^{43 ]} for a proof of concept, an agreement was reached regarding the presence of nαIFX in four out of five clinical samples (Table 1 ). It is encouraging to compare the similar figures achieved herein with those achieved by ELISA. Nevertheless, in samples nr. 1, 3, and 5, the approximate concentrations determined by ELISA were higher than those defined by SBFIA. According to the current results, LFA and ELISA have been commonly reported to exhibit moderate‐to‐good assay agreement for detecting the same analyte.^{[ 44} , ^{45 ]} The discrepancies in this study are likely due to the fundamental differences in the nature of the two detection systems, different bioreceptors, interaction matrices, interaction times, and bioreceptor concentrations. The Anti‐Human λ‐chain ELISA uses TNF as coating, IFX as first reagent, serum (with putative αIFX) as sample, and an Anti‐Human λ chain as detection antibody.^{[ 37} , ^{43 ]} This ELISA format exploits the difference in light chains that exist between IFX (κ chain) and αIFX (most of them λ chain) to detect positive samples. Noteworthy, this assay (like almost all available ELISA formats) cannot discriminate between the nαIFX and bαIFX (instead counts the overall αIFX), thus potentially resulting in higher determined values than SBFIA (which is specific to nαIFX).

Table 1.

Detection results (mean ± SD) of αIFX in human sera with the proposed assay compared to ELISA. (n = 3)

Sample nr.	ELISA [µgeq mL−1] a)	Proposed assay [µgeq mL−1] a)
1	5.6 ± 0.1	4.2 ± 0.5
2	0 ± 0.1	2 ± 0.3
3	10.9 ± 0.3	9.1 ± 0.7
4	2.4 ± 0.2	4.8 ± 0.8
5	7.7 ± 0.4	5.6 ± 0.6

^a) αIFX are expressed as µg_eq mL⁻¹, highlighting that the nαIFX used to calibrate assays differ from those produced by the patients.

Conversely, in samples nr. 4 and nr. 2, the ELISA test showed lower values than those determined by SBFIA. It is rather apparent that in both cases, the values stand at the lower side of the interval, suggesting that these patients might be in the early stages of nαIFX development; hence, the drug is masking the nαIFX presence in the performed ELISA. Notably, ELISA can only detect free αIFX, overlooking the drug‐bound antibodies and leading to potentially lower values compared to SBFIA (which detects the free and bound as complexes). Previous reports on immunogenicity models for mAb‐treated patients have indeed emphasized the existence of complexes of ADA‐drug in the early stages.^{[ 10 ]} Another hypothesis for these discordant results between the two assays might be the incubation time. ELISA requires hour‐long incubation steps, which might favor the dissociation between the low‐affinity αIFX in serum and capture/detection antibodies of the test, leading to lower values than SBFIA. Earlier studies have reported this by comparing ELISA to quicker assays (SPR) for detecting αIFX in patients.^{[ 46 ]} From an immunological perspective, it is quite possible that in the early stages of immunogenicity appearance, low‐affinity antibodies are produced by the host and might be missed in the ELISA diagnosis. A clinical study with a larger sample cohort is being conducted to explore the potential of the drug‐tolerant SBFIA in uncovering the immunogenic response earlier. MSD and ELISA drug‐tolerant assay formats have been recently reported to uncover the response earlier.^{[ 11} , ^{12 ]} Altogether, these results suggest that SBFIA could detect nαIFX in diluted human serum, and considering its POC advantages, it has remarkable potential for clinical deployment.

3. Conclusion

In the present work, a new LFA platform concept, namely SBFIA, was developed for nαIFX antibody determination. The novel test utilizes both direct and competitive binding in chromatography on a strip‐like setup to detect the analyte of interest. A new capture‐filtration analytical membrane based on affinity was conceptualized and proved effective in “filtering out” the compromising immunoreagents. The platform was optimized for detecting nαIFX and showed a threshold sensitivity of 103 ng mL⁻¹. The test was able to tolerate up to 20 µg mL⁻¹ of IFX without significant loss of signal. Furthermore, SBFIA could discern between nαIFX and bαIFX with high specificity. This paper‐based assay circumvents technical drawbacks, such as drug sensitivity and the inability to evaluate the total neutralization activity, that most of the currently used techniques face, without the need for tedious sample processing. It displays complex problem‐solving capabilities yet retains simplicity, accuracy, portability, rapidness, and cost‐effectiveness. Considering these advantages, the platform could be utilized to screen mAb‐treated patients for immunogenicity and potentially uncover the loss of response earlier. Larger trials are therefore required to provide more definitive evidence on the latter. We configured the setup for IFX and nαIFX, but by just changing bioreceptors, this assay is easily tunable to different mAb and respective ADA, as well as monitoring the adaptive immunological response, thus projecting an array of applications.

4. Experimental Section

Reagents and Immunoreagents

Tris‐buffered saline (TBS) tablets, phosphate‐buffered saline (PBS) tablets, Bovine serum albumin (BSA), Sodium chloride, Lactose, Tween 20, Triton X‐100, Freund's complete and incomplete adjuvants, polyethylene glycol Hybri‐Max, hypoxanthine‐aminopterin‐thymidine (50×) Hybri‐Max, hypoxanthine‐thymidine (50×) Hybri‐Max, 3,3' ‐Diaminobenzidine (DAB) were purchased from Sigma–Aldrich (Israel). Dulbecco's modified Eagle Medium, fetal bovine serum, penicillin‐streptomycin, L‐glutamine, HEPES buffer, sodium pyruvate, and non‐essential amino acids solutions were purchased from Sartorius (Israel). e‐Myco medium mycoplasma PCR detection kit (2.0) was purchased from LifeGene (Israel). Milli‐Q ultrafiltered H₂O (with a resistivity of 18.2MΩ cm at 25 °C) was used to freshly prepare all solutions. Human tumor necrosis factor alpha (herein TNF) (300‐01A) was purchased from Peprotech (Israel). Infliximab (Remicade) (herein IFX) was purchased from Janssen Biologics B.V. Human anti‐Infliximab non‐neutralizing antibody conjugated to horse radish peroxidase (HCA216P) (herein marker/Reporter), human anti‐infliximab neutralizing antibody (HCA233) (herein nαIFX), human anti‐infliximab binding non‐neutralizing antibody (HCA215) (herein bαIFX_a1), and human binding anti‐infliximab monovalent fragment antibody Fab’ (HCA214) (herein bαIFX_a2) were purchased from Bio‐Rad/Enco (Israel). Goat anti‐human IgG Fc (herein AHFc) antibody (I2136) was purchased from Sigma–Aldrich (Israel). Goat anti‐mouse IgG HRP (Zy‐626520) and goat anti‐mouse IgG HRP (L+H) (HP‐03) (MA1‐10371) were purchased from Rhenium (Israel).

Pads, Plasticware, and Equipment

SP (#8951) and STP (#8980) were purchased from Ahlstrom‐Munksjö (Finland). The backing card (#L‐H50), CP (#PT‐R5), and absorbent pad (#AP‐080) were purchased from Advanced Microdevices (India). UniSart NC (#CN95) was purchased from Sartorius (Israel). An automated dispenser (Easy jet LPM‐02) and a programmable guillotine (M‐70) from Advanced Microdevices (India) were used to dispense the protein lines and cut the strips, respectively. Flasks, Stericup vacuum filters, cell scrapers, and microplates were all purchased from Merck, Sigma–Aldrich (Israel). BioTek Synergy Mx Microplate Reader (BioTek) or Thermo Scientific Luminoskan was used to read ELISA signals. Chemiluminescent signals were captured with an Azure 400 imaging system (Azure Biosystems).

Patient and Animal Experiments

Blood samples were collected from healthy donors or patients receiving IFX therapy at the Rambam Health Care Campus. Sera was obtained by allowing blood to clot at room temperature for 20 min, followed by centrifugation at 1800 × g for 10 min. The research was approved by the Rambam Health Care Campus Advisory Committee on Human Experimentation (nr. 0075‐09‐RMB and 0052‐17‐RMB), and all research complied with national regulations for investigational devices, the ethical principles outlined in the Declaration of Helsinki, and with good clinical practice as described in the ICH‐GCP guidelines. Informed consent was obtained from all participants in the study. Mice were housed according to standard animal care, and protocols were approved by the Animal Experiment Committee of Ben‐Gurion University of the Negev (IL‐04‐02‐2021 012b17816_14).

Signal Generation, Acquisition, and Image Analysis

In all experiments, the samples were run for 5 min and afterward left to react with Luminol (60 µL) for another 5 min. Images were acquired in Azure 400 and processed on ImageJ (US National Institutes of Health) for maximized sensitivity.

General Conditions Optimization

To optimize assay performance, parameters such as RB composition, sugars and detergents content, pad materials, labeling agent, protein concentrations on NC, substrate type, and sample volumes were optimized. RB comprised TBS (50 mм), with Tween 20 (0.075%), and Triton X‐100 (0.075%). Pad materials were selected as follows: 8950 for SP, PT‐R5 for CP, 8980 for STP, and CN95 for NC. To determine the optimal label, the HCA216 antibody (used as the Reporter) conjugated to either HRP or 20 nm AuNPs was evaluated based on sensitivity for detecting IFX in the TNF line of the strips. HRP was selected over AuNP as the preferred label due to its superior sensitivity, achieving a lower detection limit of 6 ng mL⁻¹ compared to 180 ng mL⁻¹ with AuNPs. Furthermore, chemiluminescent reaction using luminol was chosen over the colorimetric DAB substrate, as it demonstrated a lower detection threshold (1 ng mL⁻¹ vs 6 ng mL⁻¹, respectively), thereby enhancing assay sensitivity. TNF (0.6 mg mL⁻¹) and AHFc (0.7 mg mL⁻¹) were chosen as the minimal concentrations that achieved optimal sensitivity; therefore, in the following experiments, such amounts were immobilized on the NC membrane. Typically, 35 µL of analyte sample and 35 µL of RB were applied in each strip. IFX capture‐filtration area on NC and Reporter concentration were found to be crucial for the assay's sensitivity; hence, they were described in more detail in the Optimization of the capture‐filtration area for noise reduction and Optimization of the conjugate dried Reporter.

Immunoreagent Immobilization and Strip Preparation Procedure

To prepare the strips, SP (0.6 cm) were soaked in an IFX‐containing solution (final concentration 2.1 µg mL⁻¹) composed of TBS (50 mм, pH 7.4 with 2% w/v lactose, 1% w/v BSA, 0.075% v/v Tween 20, and 0.075% v/v Triton X‐100), and dried at 37 °C for 1.5 h. Next, CP (1 cm) was soaked in a Reporter‐containing solution (final concentration 0.4 µg mL⁻¹) with the same buffer composition and dried under the same conditions. To dispense the protein lines in the NC membrane, TNF (0.6 mg mL⁻¹) and AHFc (0.7 mg mL⁻¹ with the final protein concentration adjusted to 1 mg mL⁻¹ by BSA addition) were prepared in a solution of PBS (10 mм, pH 7.4, with 3% v/v EtOH). Membranes were dried at 50 °C for 10 min. All pads were assembled on an adhesive backing card with 2 mm overlaps and cut to 4 mm‐wide strips (see Figure S1, Supporting Information, for the platform's architecture).

Optimization of the Capture‐Filtration Area for Noise Reduction

The assay's selectivity relies on the ability of the “capture‐filtration areas” immobilized on the analytical membrane to “filter out” the unbound and the monovalently bound Reporter, which otherwise would migrate to the AHFc and cause a false positive reaction. The “filtration area” was composed of immobilized IFX, which, besides preventing the Reporter from reaching the AHFc protein line, simultaneously serves as a control for confirming an adequate assay run. Two step‐by‐step experiments were conducted to find the optimal capture‐filtration area parameters: (I) the concentration of immobilized IFX was kept constant (0.5 mg mL⁻¹), and three area treatments were compared, each with a different number of dispensed IFX lines (one, two, and four lines, respectively), in a hurdle‐like fashion; (II) the number of lines was kept constant, and three concentrations of the immobilized IFX were tested (0.5, 1, and 2 mg mL⁻¹). In all steps, strips were prepared as described in the Immunoreagent immobilization and strip preparation procedure and were tested using Reporter only (without IFX or nαIFX).

Optimization of the Conjugate Dried Reporter

Due to the short interaction time between the incoming IFX/IFX‐nαIFX and the Reporter, the latter's concentration directly impacts the assay's sensitivity. However, in this assay, the Reporter's concentration, which achieves optimal sensitivity, was intertwined with the capacity of the capture‐filtration area to filter the amount of Reporter being used, avoiding the noise at the test line. Higher concentrations than the threshold would overflow the capture‐filtration IFX area and migrate to the AHFc, thus resulting in false positives. After determining the operational concentrations separately, 0.4, 0.8, and 1.2 µg mL⁻¹ of conjugate‐dried Reporter were tested. Strips were prepared as described in the Immunoreagent immobilization and strip preparation procedure. They were run only with the Reporter, using the parameters of the capture‐filtration area defined in Optimization of the capture‐filtration area for noise reduction. Results were presented in Figure S3 (Supporting Information).

Sensitivity of the Optimized Setup in Detecting nαIFX

In clinical practice, the consensus for an nαIFX cut‐off value in patients’ sera (based on which a clinical decision should be made) was still a hot topic, primarily due to the variability of the results obtained by different methods. However, ADA (8 µg mL⁻¹) has been reported as the threshold value in widely applied algorithms (such as TAXIT).^{[ 36} , ^{38 ]} A recently developed LFA qualitative test exhibits a cut‐off value of 1.3 µg_eq mL⁻¹ for anti‐Infliximab; however, after a tenfold sample dilution.^{[ 18 ]} In the current study, lower concentrations were tested. The assay was set up by spiking the RB with different concentrations of nαIFX (51.9, 103.9, 207.8, 415.6, 831.2, 1662.5, 3325, 6650, and 13 300 ng mL⁻¹). Strips were prepared as described in the Immunoreagent immobilization and strip preparation procedure, with optimized parameters (final Reporter concentration of 0.4 µg mL⁻¹ and three IFX capture lines, each 2.4 mg mL⁻¹). Measurements were performed as described in the Signal generation, acquisition, and data analysis. The same optimized parameters were followed in the subsequent experiments.

Drug‐Tolerance Tests of the Optimized Setup

Drug tolerance determines the feasibility of the setup's real‐life application. In clinical practice, standard guidelines define serum IFX levels between 3–7 µg mL⁻¹ as therapeutic, < 3 µg mL⁻¹ subtherapeutic, and > 7 µg mL⁻¹ supratherapeutic.^{[ 36} , ^{38 ]} To test the present assay's drug‐tolerance capability in these ranges, in addition to the inherent, pre‐dried IFX that was used to saturate samples nαIFX, the RB was spiked with different concentrations of sample IFX (625, 1250, 2500, 5000, 10 000, and 20 000 ng mL⁻¹). Optimized strips were prepared as described in the Immunoreagent immobilization and strip preparation procedure. Samples’ nαIFX were kept at 2.4 µg mL⁻¹ among treatments. Measurements were performed as mentioned in Signal generation, acquisition, and data analysis.

Specificity of the Optimized Setup

To test the platform's specificity for nαIFX, different treatments were prepared as follows: sample spiked with nαIFX as positive control, sample with a whole antibody derived from human cells (bαIFX_a1), sample with a fragment antibody derived from human cells (bαIFX_a2), sample with an antibody derived from murine hybridomas (bαIFX_a3) and a sample with a mixture of all the above. The concentration of each antibody in the above treatments was maintained at 2.2 µg mL⁻¹. Optimized strips were prepared as described in the Immunoreagent immobilization and strip preparation procedure, and measurements were performed as mentioned in Signal generation, acquisition, and data analysis. bαIFX_a3 was produced using hybridoma technology as previously described,^{[ 47 ]} with reported modifications.^{[ 48 ]} Briefly, 7‐week‐old BALB/c female mice were immunized subcutaneously with IFX and later screened with an indirect ELISA for anti‐IFX antibodies in murine sera (see Figure S4a, Supporting Information). After fusion, the same ELISA format was applied to screen for secreted antibodies in hybridoma supernatants (Figure S4b, Supporting Information). Next, a competitive ELISA was used to select non‐neutralizing binders (Figure S5, Supporting Information). The strong binders were selected for specificity testing with the strip setup. Strips were prepared and tested using standard procedures.

Human Serum Analysis

Due to the AHFc line being the capturing element for the analyte, a cross‐interference from serum IgG was suspected. To test the putative interference and serum effect, naive human sera were diluted ten times with RB and compared to pure RB, both spiked with 6600 ng mL⁻¹ nαIFX. Measurements were performed as described above. Further, after optimization of the AHFc line, sera: running buffer solution was spiked with concentrations ranging from 0 to 3325 ng mL⁻¹. Optimized strips were prepared, and the measurements were performed as described above. Next, to demonstrate the applicability of the proposed assay in real samples, for a proof of concept, sera collected from Crohn's patients showing clinical signs of response loss to IFX were analyzed (five samples). First, the serum was fractionated and tested by λ chain ELISA performed as previously described,^{[ 37} , ^{43 ]} for the presence of αIFX antibodies. Afterward, serum samples were diluted ten times in the RB to test the samples with the proposed SBFIA. Results were compared to assess concordance.

Statistical Analysis

Signal intensities were presented as mean ± SD, considering three repetitions (n = 3). Two‐tailed Kruskal‐Wallis (with Dunn's multiple comparisons) and Mann‐Whitney tests (as applicable) were used to compare treatments with the significance threshold set at 0.05. Plotting and statistical tests were performed using GraphPad Prism v. 8.0.0 (San Diego, CA, USA) and OriginPro v. 2024b (Northampton, MA, USA). P values were presented with ^* at the 0.05 level, ^** at the 0.005 level, and ^*** at the 0.0005 level.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Kokojka F., Ramon R., Pressman S., Chowers Y., and Marks R. S., “Drug‐Tolerant, Chemiluminescent Lateral Flow Immunoassay Platform for the Determination of Neutralizing Anti‐Drug Antibodies.” Small 21, no. 40 (2025): e05975. 10.1002/smll.202505975

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.