1. Introduction
Anti-tumor necrosis factor (TNF) inhibitors have revolutionized the management and prognosis of various inflammatory diseases [
1]. However, some patients initially resist treatment, or experience a decline in efficacy over time. For example, a previous study reported the failure of adalimumab, a monoclonal human antibody, to maintain adequate drug concentrations in secondary non-responders, with the potential formation of antibodies against the drug [
2]. Consequently, assessing drug concentrations and detecting anti-drug antibodies in patients with suboptimal drug responses may offer a more precise understanding of their condition and guide subsequent treatment strategies [
3]. The measurement of drug concentrations and antibodies is predominantly conducted using enzyme-linked immunosorbent assay (ELISA) methods, which, despite their high validity, are constrained by the necessity of pooling sera from multiple patients, the requirement for kit calibration, and the extended time needed to obtain results, often rendering them impractical for same-day clinical decision-making [
4]. Given these shortcomings of ELISA, point-of-care (POC) medical technologies are gaining popularity because of the increasing trend toward individualized medical services, where the prompt application of assessment results enhances time and cost efficiency. The AFIAS platform represents a promising POC solution, utilizing automated fluorescence-based detection with cartridge-based assays for both drug concentration and anti-drug antibody measurements.
This study aimed to assess the analytical performance and clinical utility of the AFIAS automated fluorescence-based point-of-care (POC) immunoassay system, employing the AFIAS-10 device with adalimumab and anti-adalimumab antibody cartridges, compared with established ELISA methods. Clinical factors associated with low drug concentrations and antibody emergence were also evaluated.
3. Discussion
In this study, the newly developed lateral-flow POC cartridge for measuring serum adalimumab concentration and anti-drug antibodies demonstrated clinically applicable levels of accuracy and reliability compared with those of existing commercially available ELISA devices. The difference in adalimumab concentration was acceptable for clinical utility based on the Bland–Altman plot of difference, as the outlier was plotted at a higher mean concentration than the usual therapeutic cut-off value of 5–12 μg/mL [
5,
6,
7]. The detection of AAAs was largely consistent across the three kits, although some patients exhibited varying results between kits, a phenomenon that was confirmed by the commercially available kits. The POC cartridge also exhibited consistent characteristics with respect to patient clinical profiles, rendering it suitable for clinical application. The presence of ANAs correlates with patient immunological characteristics, is more prevalent in women, and is associated with autoimmune diseases [
8]. In this study, positive ANA results were pre-tested in all patients and were not related to the use of TNF inhibitors. In univariate analysis, ANA was found to be associated with AAA development. However, this association was not evident in multivariate analysis, with a small margin of difference. This discrepancy may be attributed to the limited sample size of the patient cohort. These findings indicate that baseline ANA positivity is associated with AAA development, which is consistent with the findings of previous research. Indeed, previous studies have shown that patients with baseline ANAs and higher titers prior to commencing TNF inhibitors were more likely to develop anti-drug antibodies [
8]. In addition, lower drug concentrations were observed in patients with longer intervals between injections or higher BMI [
2,
9]. This expands beyond prior work that evaluated the AFIAS platform with one infliximab and one adalimumab ELISA in patients with various rheumatic diseases [
10]. Our comprehensive analytical validation employs multiple agreement metrics, providing deeper insight into analytical performance. Furthermore, we extend the clinical utility evaluation by examining patient-specific factors associated with adalimumab trough levels and anti-adalimumab antibody positivity through multivariate analyses, addressing knowledge gaps.
Current methods measure the levels of free anti-drug antibodies but do not detect the bound form of anti-drug antibodies. This implies that certain antibodies may not be detected because they bind to the drug, even though the patient has developed AAA. Notably, this phenomenon is likely to occur in environments with high drug concentrations. This explains why AAA is rarely observed in patients with elevated drug levels, yet it is detected in those with lower concentrations with current method.
An important limitation of the AFIAS method is the detection of only free anti-adalimumab antibodies, whereas drug-tolerant assays can detect both free and bound AAA. This limitation has significant clinical implications for therapeutic decision-making. Recent studies have demonstrated that drug-tolerant assays detect AAA in 63–90% of patients compared to 21–30% using drug-sensitive methods such as AFIAS [
11,
12,
13]. In patients with high circulating adalimumab levels, free AAA may be undetectable despite the presence of clinically relevant bound antibodies that form immune complexes. These immune complexes can lead to increased drug clearance and predict future treatment failure, even when trough drug levels appear therapeutic [
2]. The clinical consequence of this limitation is the potential for false-negative AAA results in patients who may benefit from treatment modification. Patients with bound AAA detected by drug-tolerant assays show increased risk of treatment failure within 6–12 months, even when free AAA assays remain negative [
11,
14]. This suggests that AFIAS results should be interpreted with caution, particularly in patients with adequate drug levels but suboptimal clinical response. For optimal clinical decision-making, the combination of free AAA measurement (as provided by AFIAS) with drug concentration assessment may help identify at-risk patients. However, clinicians should be aware that negative free AAA results do not exclude the presence of clinically relevant immunogenicity, particularly in patients with sustained high drug levels or those being considered for dose reduction.
In general, the detection of anti-drug antibodies can be performed using various assay methodologies, including radioimmunoassays, ELISA, and cell-based reporter gene assays, each with distinct characteristics. The choice of assay significantly influences the types of antibodies detected and their clinical interpretation [
4,
15,
16]. Among them, specially designed drug tolerance assays can detect both free and binding forms of AAAs but require an additional step to isolate binding anti-drug antibodies, which increases the cost of testing [
12,
16,
17]. However, the additional benefits and utility of measuring total anti-drug antibodies compared to free antibodies remain unclear [
18]. Furthermore, no standardized and validated method is currently available to distinguish between more potent and weak neutralizing antibodies, regardless of the binding or free form of AAAs [
13,
19,
20]. Therefore, the measurement of both free and binding AAAs does not necessarily explain all cases in which treatment remains effective or has reduced effectiveness [
21]. However, in patients with reduced drug effectiveness, detection of reduced drug concentrations and anti-drug antibodies may guide the modification of treatment strategies. For instance, high drug concentrations and the absence of anti-drug antibodies, coupled with high disease activity, suggest an inflammatory response pathophysiology in which TNF inhibitors have little efficacy, warranting consideration of switching to biologics with a different mechanism or synthetic-targeted disease-modifying anti-rheumatic drugs [
4]. Conversely, if the drug concentration is low and the antibodies are not detected, adjusting the dose or dosing interval may be appropriate. In some cases, remission can be achieved with a low drug concentration, absence of anti-drug antibodies, and low disease activity [
22]. However, it remains unclear whether the trough adalimumab concentration and AAA detection warrant the modification or switching of therapeutic modalities to maintain achievable disease control. Some reports have shown that trough drug concentration and detection of anti-drug antibodies have poor predictability of disease activity in the real world [
3,
23].
This analysis revealed some differences between the measuring devices, but these differences were within clinically acceptable ranges and were not specifically attributable to lower POC device specificity, as the differences were also observed among commercially available kits. However, the observed discrepancy, particularly at supratherapeutic concentration levels, which was measured to be larger with POC devices, should be taken into account before deciding to reduce the dosage or extend the injection interval.
The clinical implementation of point-of-care (POC) therapeutic drug monitoring represents a paradigm shift from traditional laboratory-based approaches in rheumatology practice. Recent evidence from inflammatory bowel disease demonstrates that proactive TDM significantly improves clinical outcomes and enables efficient drug de-escalation in patients achieving remission [
24]. The AFIAS platform’s 10–12 min turnaround time addresses a critical workflow limitation, as conventional ELISA-based TDM requires 1–3 weeks for results, often necessitating multiple patient visits and delayed treatment adjustments [
25,
26]. This rapid availability enables same-day clinical decision-making, potentially reducing follow-up visits, treatment delays, and overall healthcare costs, though specific cost-reduction percentages require validation in rheumatology populations. The integration of POC TDM into outpatient settings requires consideration of staff training, quality management systems, and regulatory compliance, but offers the potential to transform reactive, empirically guided treatment approaches into proactive, data-driven therapeutic optimization [
27]. Future integration with electronic health records and clinical decision support systems may further enhance the clinical utility of POC TDM platforms, supporting the development of personalized biologic therapy algorithms and advancing precision medicine in rheumatology practice.
In the context of real-world clinical practice, the measurement of drug concentration or the detection of anti-drug antibodies should not be interpreted as a single binary decision-making tool. As demonstrated by the longitudinal drug retention results in this study, a low drug concentration and the presence of anti-drug antibodies are indeed associated with an increased likelihood of drug discontinuation and the transition to alternative agents. However, this process does not occur within a single time frame; rather, it exhibits a gradual trajectory, indicating a higher probability of eventual drug retention failure. In cases where a patient is in deep remission, the immediate likelihood of reactivation is minimal, even in the presence of low drug concentrations and anti-drug antibodies. However, the potential for future reactivation remains. This underscores the importance of evaluating cost-effective strategies to identify individuals who are most likely to benefit from these assessments. We illustrate this implication in
Figure 6.
This study has some limitations. First, the study had a relatively small sample size. Although more exceptional cases might emerge in a larger cohort, it remains challenging to ascertain whether the observed bias is unique to the patients in this study. Second, the clinical significance of measuring the total anti-drug antibodies remains unclear. This method measures free anti-drug antibodies; however, detecting these antibodies becomes challenging as the drug concentration increases compared to the total anti-drug antibody measurement. However, the measurement of total anti-drug-antibody levels does not confirm the specific functional aspect of the neutralizing degree, further complicating its clinical relevance. Third, the study may be subject to potential selection bias due to recruiting from a single tertiary referral center, which could impact the clinical relevance. Further research is required to establish the cost-effectiveness and to provide more detailed guidance on clinical workflows.
In conclusion, this study demonstrates that POC measurement of adalimumab concentrations and anti-drug antibodies using the lateral-flow cartridge exhibits clinical significance and reliability comparable to those of existing commercially available ELISA kits. The results correlated well with the disease activity and clinical characteristics of patients, suggesting potential clinical relevance. Nonetheless, conducting larger, multicenter studies is essential to validate the observed associations and confirm the device’s applicability across diverse patient populations.
4. Materials and Methods
4.1. Study Design and Population
Ninety-six patients receiving adalimumab injections and attending the Department of Rheumatology at Severance Hospital, a tertiary hospital in Seoul, South Korea, were consecutively recruited based on their order of visits between 27 August 2023 and 14 October 2023. The clinical characteristics of the patients were retrieved from the medical records, and baseline data for ANA and RF were retrospectively obtained from the pre-adalimumab treatment records. Disease activity was assessed using the standard disease activity scale for each diagnosis and categorized arbitrarily by the investigator as none (remission), mild, moderate, or severe. One year following the commencement of enrollment, we assessed the current status of ADA continuation within the study populations.
4.2. Sample Collection and Measurements
Serum samples were collected on the day of the visit to measure anti-drug antibodies and drug concentrations, including inflammation levels. Three commercially available ELISA methods (RIDASCREEN® ADM Monitoring [R-Biopharm AG, Germany], IDKmonitor® Adalimumab Drug Level [Immundiagnostik AG, Germany], and LISA-TRACKER Duo Adalimumab [Theradiag, France]) were used to measure adalimumab concentration, and two ELISA kits (IDKmonitor® Adalimumab free ADA [Immundiagnostik AG, Germany] and LISA-TRACKER Duo Adalimumab [Theradiag, France]) were used for AAA detection. ELISA was performed according to the manufacturer’s protocols, ensuring consistency in the assay procedures. The results were compared with the measurements obtained using the commercially available AFIAS Adalimumab and AFIAS Free Anti-Adalimumab (Boditech Med Inc., Chuncheon, Republic of Korea).
4.3. AFIAS POC Assays and Reagents
The AFIAS assays (AFIAS Adalimumab and AFIAS Free Anti-Adalimumab) were conducted using the AFIAS-10 analyzer (Boditech Med Inc., Chuncheon, Republic of Korea), an automated, integrated system designed to efficiently process samples from input to result output, ensuring ease of use and high-throughput capacity. The platform enables the simultaneous execution of up to 10 tests without manual intervention. For each test, 100 µL of the sample was loaded into the sample well, and the system automatically processed the assay, including incubation, detection, and interpretation of the results. The operation and basic principles of the AFIAS are based on fluorescence-based lateral flow immunoassay. AFIAS Adalimumab assay results were obtained after 10 min, while AFIAS Free Anti-Adalimumab results were available after 12 min. The measurement ranges for both the AFIAS and ELISA are summarized in
Supplementary Tables S8 and S9.
The assays utilize disposable all-in-one cartridges containing fluorescence-labeled antibodies, buffers, and test strips, which enable fully automated sample-to-result processing without manual reagent preparation. Quality control materials supplied by the manufacturer were used according to protocol. Additionally, the World Health Organization (WHO) international standard material for Adalimumab (NIBSC code 17/236) was employed as a reference to validate the assay.
4.4. Analytical Validation
Analytical validation of the AFIAS Adalimumab assay was performed according to CLSI guidelines. Precision assessment, including repeatability, within-laboratory precision, and lot-to-lot reproducibility, was evaluated at four concentrations (2, 5, 8, and 20 µg/mL) using three reagent lots. Replicate measurements were conducted to calculate coefficients of variation (CV) and recovery rates based on the WHO international standard (
Supplementary Tables S1 and S2). To assess inter-operator variability and minimize potential bias, two trained laboratory technicians independently performed all AFIAS and ELISA measurements under standardized protocols. Inter-operator precision was analyzed by comparing replicate results (
Supplementary Table S1).
4.5. Statistical Analysis
For baseline and proportional bias between the measurement methods, both the Passing–Bablok analysis and Bland–Altman plots were used. Passing–Bablok regression was used to determine the intercept and slope of the linear regression using the equation y = a + bx [
28]. To assess the agreement between the two methods in quantification, the Bland–Altman analysis was performed by examining the mean difference and percent mean difference, establishing limits of agreement (LOAs) [
29]. LOAs were calculated as the mean difference ± 1.96 times the standard deviation of the differences. The agreement was sufficient if the concentration difference was within the LOA for 67% of the sample pairs [
29]. Lin’s concordance correlation coefficient (ρ
c) was used to measure agreement or reliability between two sets of continuous measurements. The utility of ρ
c lies in its ability to capture both the precision (Pearson’s correlation coefficient) and accuracy (Cβ) of the agreement between the two sets of agreements. Cβ is a bias that considers both the systematic difference (bias) and dispersion (variability) between measurements. Altman considered the interpretation of ρ
c > 0.8 to be excellent, while McBride indicated that <0.9 suggests poor agreement [
30]. Intraclass correlation coefficients (ICCs) were calculated for the adalimumab concentration values using a two-way mixed-effects model to investigate the consistency of the results between our method and commercially available ELISA kits. ICC is a measure of reliability and ranges from 0 to 1. Generally, it describes the ratio of variance of interest (i.e., due to patient differences) to the total variance (i.e., variance of interest + noise). For example, if the unwanted variance (noise) is equal to or larger than the variance due to patient differences, the reliability is considered poor, and the ICC will be ≤0.5. In contrast, ICC values above 0.8 or 0.9 are often regarded as indicators of good or excellent reliability [
30,
31]. To examine the dichotomous determination of AAA, the percent agreement was calculated. Cohen’s or Fleiss’s kappa statistics were calculated for decision agreements between two or more than three groups to detect AAAs [
28]. The kappa value was interpreted according to the criteria proposed by Landis et al. [
32]: k < 0, no agreement; k (0.00–0.20), slight agreement; k (0.21–0.40), fair agreement; k (0.41–0.60), moderate agreement; k (0.61–0.80), substantial agreement; and k (0.81–1.00): almost perfect agreement.
Regarding the necessary sample count for calculating ICC, a sample size of 53 participants with two observations per participant achieved 80% power to detect an ICC of 0.5 under the alternative hypothesis when the intraclass correlation under the null hypothesis was 0.2, using an F-test with a significance level of 0.05. For kappa, in a test for agreement between two raters using the kappa statistic, a sample size of 53 participants achieved 80% power to detect a true kappa value of 0.7 in a test of H0: kappa = κ0 vs. H1: kappa ≠ κ0 when there were three categories with frequencies equal to 0.2, 0.3, and 0.5. The power calculation was based on a significance level of p < 0.05.
Regarding clinical relevance, we analyzed the associations of drug concentrations and anti-drug antibody detection with clinical characteristics. The chi-square test or Fisher’s exact test was performed as appropriate to compare categorical variables. For normally distributed continuous variables, a t-test was used; otherwise, the rank-sum test or Wilcoxon signed-rank test was used for paired samples. Linear regression analysis was performed to calculate the coefficient values for each variable of adalimumab concentration, and logistic regression analysis was performed for the development of anti-drug antibodies or disease activity. Backward selection was employed to identify the pertinent variables for the development of AAA in the context of multivariate logistic analysis. In addition, to assess the longitudinal outcomes associated with the retention of ADA within the study group, we performed both univariate and multivariate Cox regression analyses. The proportional-hazard assumption was verified using Schoenfeld residuals. By employing initial ADA concentrations and AAA detection results, we evaluated the survival curve for the continuation of ADA administration. Moreover, we calculated the estimated hazard function plots over the observed failure times, categorizing the patient group based on ADA concentration and the presence of AAA. Statistical significance was set at p < 0.05.
All statistical analyses were performed using STATA for Windows version 18.0 (Stata Corp., College Station, TX, USA).
4.6. Ethics
The Institutional Review Board (IRB) of Severance Hospital approved this study (IRB: 2022-1531-001). All the patients provided informed consent for participation in the study. The study was conducted in full conformity with the current revision of the Declaration of Helsinki (2013) and in accordance with local guidelines and regulations.