Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

A competitive ligand-binding assay for the detection of neutralizing antibodies against dostarlimab (TSR-042)

A competitive ligand-binding assay for the detection of neutralizing antibodies against... Dostarlimab is a humanized anti–PD-1 monoclonal antibody. Dostarlimab (JEMPERLI; TSR-042) was recently approved in the USA and in the EU. The presence of neutralizing antibodies (NAbs) is a cause for concern because they block the therapeutic function of the antibody and reduce drug efficacy. Therefore, programs developing therapeutic biologics need to develop and validate assays that adequately assess the presence of NAbs in the serum of patients treated with biologic therapies. Presented here is the development and validation of a competitive ligand-binding assay that specifically detects anti-dostarlimab NAbs in human serum. Precision, sensitivity, hook effect, selectivity, assay robustness, stabilities, and system suitability were evaluated. In addition, drug tolerance of the assay with the implementation of a drug removal process was investigated. The cut point factor for the detection of NAbs in human serum at a 1% false-positive rate was determined. The assay’s precision, sensitivity, hook effect, selectivity, robustness, and drug interference were tested and found to be acceptable. With system suitability and stability established, this assay has been used to evaluate NAbs to guide the development of dostarlimab. Trial registration: Clinicaltrials.gov, NCT02 715284. Registered 9 March 2016 Keywords: Dostarlimab, Endometrial cancer, Neutralizing antibodies, Immunotherapy, Competitive ligand-binding assay Introduction monoclonal antibody (mAb) (Oaknin et  al. 2018). In the Immunotherapy with antibodies directed at the pro- USA, dostarlimab was recently approved as a mono- grammed death 1 (PD-1)/PD ligand 1(PD-L1) axis has therapy in adult patients with mismatch repair-deficient shown remarkable activity in a variety of tumor types (dMMR) recurrent or advanced endometrial cancer that in clinical trials and has been approved worldwide for has progressed on or after a platinum-containing regi- a number of malignancies (Robert et  al. 2014; Weber men or dMMR recurrent or advanced solid tumors that et al. 2015; Fehrenbacher et al. 2016; Antonia et al. 2017; have progressed on or following prior treatment and Apolo et  al. 2017; Seiwert et  al. 2016; Ansell et  al. 2015; who have no satisfactory alternative treatment options El-Khoueiry et  al. 2017; Kaufman et  al. 2016; Gong (US Food and Drug Administration 2021a; US Food et  al. 2018; Ahn et  al. 2019; Khozin et  al. 2019). Dostar- and Drug Administration 2021b; Andre et  al. 2021; limab (JEMPERLI; TSR-042) is a humanized anti–PD-1 ClinicalTrials.gov 2016). In the EU, dostarlimab was recently approved as a monotherapy in adult patients with recurrent or advanced dMMR/microsatellite insta- bility–high endometrial cancer that has progressed on *Correspondence: slu@scholarrock.com GlaxoSmithKline, 1000 Winter Street, Waltham, MA 02451, USA or after treatment with a platinum-containing regimen Full list of author information is available at the end of the article (Oaknin et al. 2018; Andre et al. 2021; ClinicalTrials.gov Weiping Shao was employed by Frontage Laboratories when the 2016; European Medicines Agency 2021). As with other research was conducted. © The Author(s) 2021. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. Zhang et al. AAPS Open (2021) 7:8 Page 2 of 14 immunotherapeutic agents, anti-drug antibodies (ADAs) Medicines Agency recognize both cell-based and non– may emerge and be detectable after administration of cell-based competitive ligand-binding assays as valid dostarlimab (Pineda et al. 2016). However, dostarlimab is measurements of NAbs in some cases (Jolicoeur and a humanized mAb, and as such, has a lower risk of elic- Tacey 2012; Wu et al. 2016). The selection of either cell- iting ADAs or neutralizing (NAbs) than non-humanized based or non–cell-based assays is driven by several dif- mAbs (Davda et al. 2019). ferent variables, including therapeutic mechanism of However, the production of ADAs is generally a treat- action (MOA), assay performance characteristics, and ment risk with therapeutic antibodies and can lead to an risk of immunogenicity, with therapeutic MOA being altered pharmacokinetic (PK) profile, which in turn may the primary determinant (Wu et al. 2016). impact the safety and efficacy of therapeutic antibodies Here, we present the development and validation of a (Garcês and Demengeot 2018; Bloem et al. 2017). ADAs competitive ligand-binding assay that is specific for the can be (1) binding, leading to minimal or no impact; (2) detection of anti-dostarlimab NAbs in human serum. We clearing, leading to impact through an altered PK pro- validated the assay’s precision, sensitivity, hook effect, file; (3) sustaining, leading to a prolonged exposure that selectivity, robustness, stability, and system suitability. In may change the efficacy and/or safety; or (4) neutralizing, addition, we investigated the drug tolerance of the assay leading to reduced pharmacological activity of the thera- with the implementation of a drug removal process, as peutic antibody (Garcês and Demengeot 2018; Bloem well as the possibilities for false-negative results. The util - et al. 2017). ity of this assay as it pertains to the dostarlimab clinical u Th s, the presence of NAbs is concerning because they development program is also briefly discussed. block the therapeutic function of the antibody and are likely to lead to reduced efficacy (Garcês and Demengeot Materials and methods 2018; Bloem et  al. 2017; Gunn et  al. 2016). Further- Materials more, NAbs may lead to safety issues, such as hypersen- Critical reagents for this assay were as follows: a mouse sitivity and anaphylaxis Gunn et  al. 2016; Krishna and anti-dostarlimab mAb (Precision Antibodies, Columbia, Nadler 2016). Therefore, programs developing therapeu - MD) generated from clone #6G10 at a concentration of tic antibodies need to develop and validate assays that 1.15 mg/mL; dostarlimab (WuXi, Philadelphia, PA) at a adequately assess the presence of NAbs in the serum of concentration of 20.7 mg/mL; biotinylated human PD-1 patients treated with such therapeutic agents. (ACRO Biosystems, Newark, DE) at a concentration of Cell-based reporter assays have been developed and 200 μg/mL; and SULFO-TAG–labeled human PD-L1 validated for detecting NAbs, including a cell-based bio- (CR0; Meso Scale Diagnostics [MSD], Rockville, MD) at luminescent reporter assay that quantifies therapeutic a concentration of 2106 μg/mL. mAbs against PD-1 and PD-L1 (Liao et  al. 2012; Cong et al. 2015; Jolicoeur and Tacey 2012). Preparation of dostarlimab drug solution Regulatory agencies prefer cell-based assays because The initial stock solution of dostarlimab was diluted they rely on cellular responses to NAb-mediated inhi- in LowCross-Buffer (Boca Scientific, Westwood, MA) bition of therapeutic antibodies (Wu et  al. 2016). to a concentration of 1608 ng/mL, which is 6 times the Because of their mode of detection, these assays have desired final concentration (268 ng/mL). The prepared been considered more biologically relevant than non– dostarlimab solution was aliquoted and stored at − 70 °C. cell-based assays (Bloem et  al. 2017; Liao et  al. 2012; Cong et  al. 2015; Jolicoeur and Tacey 2012). However, cell-based assays are labor intensive and highly vari- Preparation of assay control samples able and exhibit low serum tolerance and poor drug The negative control (NC) was prepared by adding 250 tolerance (Bloem et  al. 2017; Liao et  al. 2012; Cong μg/mL dostarlimab (GSK, Waltham, MA) into pooled et  al. 2015; Jolicoeur and Tacey 2012; Wu et  al. 2016). normal human serum (pNHS; Bio IVT, Westbury, NY) In contrast to cell-based assays, non–cell-based assays and freezing the solution for 12 h. After thawing, dostar- often rely on binding of the drug and target for signal limab was removed from the NC samples via the drug detection and quantitation. Competitive ligand-binding removal procedure (see below). The low-positive control assays, used to characterize NAbs, have been shown (LPC) and high-positive control (HPC) were prepared to provide higher sensitivity, a wider dynamic range, by mixing dostarlimab and anti-dostarlimab NAb clone increased precision, and better matrix tolerance than 6G10 in pNHS at 250 μg/mL and 500 ng/mL, respectively, cell-based assays (Wu et  al. 2018). Although regulatory for the LPC and at 250 μg/mL and 2000 ng/mL, respec- agencies generally prefer cell-based assays, the United tively, for the HPC. The solution was then subjected to the States Food and Drug Administration and the European same freeze–thaw–drug removal procedure as the NC. Zhang  et al. AAPS Open (2021) 7:8 Page 3 of 14 Once prepared, the NC, LPC, and HPC were stored at − dark. The final drug concentration was 268 ng/mL, which 70 °C. was selected based on the following: the linearity range for dostarlimab drug concentration to ECL signal was Drug removal procedure found to be 0–300 ng/mL; the highest drug concentra- Excess dostarlimab was removed from serum and con- tion in this range led to the lowest ECL signal in the test trol samples prior to performing the NAb detection sample; and any ECL signal increase due to NAb interac- assay. Biotinylated human PD-1 stock solution (200 μg/ tion in the assay introduced a better signal to noise ratio. mL) was prepared by dissolving 200 μg of PD-1 in 1 mL Therefore, approximately 90% of the 300 ng/mL (which of dH O. Magnetic streptavidin Dynabeads (Thermo was considered to be the high limit of quantification) Fisher Scientific, Waltham, MA) were prepared by trans - was used to maintain the assay reproducibility. From ferring 140 μL of streptavidin magnetic beads to a tube each well of the dilution plate, 50 μL was transferred to (approximately 50% of bead volume) where they were the washed streptavidin-coated plate and incubated for subsequently washed once using the magnetic stand with 1 hour at ambient temperature in the dark. Following 0.5 mL of wash/bind buffer. To bind human PD-1 to the incubation, the wells were washed with Tween-20 PBS, as magnetic beads, 70 μL (14 μg) of biotinylated PD-1 pro- described above, then 100 μL of 2× Read Buffer (MSD) tein was added to the washed beads, and the mixture was was added to each well. The plate was subsequently read incubated for 1 hour at ambient temperature while shak- on an MSD SECTOR Imager (MSD) to attain the electro- ing at 1000 rpm. The PD-1 solution was then removed chemiluminescence (ECL) signal. from the beads using the magnetic stand, and the beads were washed 3 times using the magnetic stand with 0.5 Statistical methods mL wash/bind buffer. Data description The positive control (PC) was prepared by adding Sixty individual human cancer serum samples were dostarlimab and NAb to pNHS to achieve a final concen - obtained from the disease population, were divided into 4 tration of 250 μg/mL of dostarlimab and the desired con- groups with 30, 30, 40, and 20 samples in groups A, B, C, centration of NAb in 70 μL of pNHS. Drug removal was and D, respectively. The plate for group D also included accomplished by adding a one-tenth volume of 1 M ace- 11 sensitivity samples. Each group was tested twice by 2 tic acid to serum or control solutions and incubating at analysts across 3 days. Each analyst tested at least 1 set of ambient temperature for 2 h, resulting in a pH of 3.0 with sensitivity samples. The reported value for each test sam - no change of ADA affinity observed. A one-tenth volume ple was the mean ECL response from duplicate wells on of 1 M Tris base was added to neutralize the samples. a plate. The NCs were tested 4 times on each plate, and The sample solution was immediately added to the pre- the means of the duplicate wells were reported for a total washed PD-1–coated magnetic beads and incubated for 1 of 32 ECL values. The LPC and HPC samples were tested hour at ambient temperature while shaking at 1000 rpm. twice on each plate for a total of 16 mean ECL values for The samples were then separated from the streptavidin each PC. beads by magnetic separation and subsequently used in the NAb assay, or alternatively, stored at − 70 °C. Sources of variation All statistical analyses were completed using JMP Pro NAb detection assay version 12.1.0 (JMP; Cary, NC), and methods used were The streptavidin-coated Gold SECTOR plates (MSD) consistent with recommendations by Shankar et  al. were blocked with 1% bovine serum albumin (Gemini (2008). Statistical outliers were determined using a quar- Bio, West Sacramento, CA) in phosphate-buffered saline tile range outlier method. The statistical outliers in the (PBS) for 1 hour at ambient temperature with shaking upper and lower quartiles were defined as Q3 + 1.5 × at 450 rpm and thoroughly washed (3 cycles on a plate (Q3 − Q1) and Q1 − 1.5 × (Q3 − Q1), where Q3 is the washer with 0.05% Tween-20 in PBS). In a dilution plate, 75th percentile and Q1 is the 25th percentile. Outliers 20 μl of prepared dostarlimab solution (1608 ng/mL) was identified by these criteria were removed, and the model added to 60 μL of drug removal samples or controls, and was re-fit. Normality of residual values was evaluated by the resulting solutions were incubated for 1 hour at ambi- the Shapiro-Wilk test (Mire-Sluis et  al. 2004). In addi- ent temperature with shaking at 450 rpm. A Master Mix tion, the skewness coefficient was calculated as a relative solution of 12 nM biotinylated PD-1 and 12 nM SULFO- measure of symmetry (Shapiro and Wilk 1965). TAG–labeled human PD-L1 was prepared in Low Cross- If skewness coefficient was less than − 1 or greater than Buffer. Then, 40 μL of Master Mix was added to each + 1, the distribution of the data was highly skewed; if well of the dilution plate, which was incubated for 1 hour skewness was between − 1 and − 1/2 or between + 1/2 at ambient temperature with shaking at 450 rpm in the and + 1, the distribution was moderately skewed; and if Zhang et al. AAPS Open (2021) 7:8 Page 4 of 14 desired NAb concentrations that span the cut point into skewness was between − 1/2 and + 1/2, the distribution pNHS with dostarlimab at 250 μg/mL. The sensitivity was approximately symmetric. The parametric cut point samples underwent the drug removal treatment prior to estimates are typically recommended if normality dis- analysis. Six runs were performed by 2 analysts over 3 tribution is confirmed (if P > 0.05); however, if normal - separate days. ity cannot be assumed, the parametric cut point estimate may be used if the data are not highly skewed (if skew- ness is between − 1 and + 1). Assay precision Intra-assay precision was evaluated by running 6 inde- NAb assay cut point pendent validation samples for NCs, LPCs (533.6 ng/ Signal to noise (S/N) ratio values were obtained by divid- mL), and HPCs (2000 ng/mL) on a single plate. The inter- ing the ECL signal value from each individual serum assay precision was measured using the PCs and NCs in sample by the mean of the NC ECL values from the cor- all accepted validation runs. responding plate. After exclusion of analytic outliers from the set of NC values, the remaining S/N ratio values were Hook effect used for statistical analyses. Homogeneity of inter-plate To determine if the assay demonstrated a hook effect, sample variances across the 8 plates was investigated per an ultrahigh concentration (20,000 ng/mL) of NAb was Levene’s test using S/N ratio values (Levene 1960). prepared in pNHS. Serial 2-fold dilutions of this sample in pNHS generated samples at 10,000, 5000, 2500, 1250, Cut point factor determination 625, 312.5, 156.25, 78.13, and 39.06 ng/mL, which were Floating cut point factors (CPFs) for parametric and evaluated as replicates. non-parametric screening were determined using the S/N ratio values. A parametric method was used to cal- Selectivity culate robust estimates of the mean and standard devia- Ten individual cancer serum samples, 5 2% hemolyzed tion (SD) of all S/N ratio values. The parametric 1% error normal serum samples and 5 lipemic normal serum sam- rate floating CPF was then determined by multiplying the ples (450 μg/dL) containing dostarlimab at 250 μg/mL SD value by the 99th quartile of the t-distribution (with (a) unspiked and (b) spiked with anti-dostarlimab NAb degrees of freedom equal to the number of S/N ratio val- at the LPC level (533.6 ng/mL), underwent drug removal ues minus 1) and adding the product to the mean of S/N treatment and were tested. ratio values. The non-parametric 1% error rate CPF was determined by calculating the empirical 99th percentile Drug interference of S/N ratio values. Interference in the detection of LPC was evaluated by using the isotype IgG4 antibody TSR-022, which is an False‑negative error rate anti-TIM3 antibody. The results were considered accept - The false-negative error rate, defined as the probability of able if the TSR-022–spiked LPC sample values were observing an S/N ratio less than the CPF in a known pos- within their respective ranges, which were established itive sample, was calculated based on the mean and SD of during this validation. Drug interference was tested using all LPC S/N ratio values and a t-distribution with degrees 3 sets of LPC and NC (a) unspiked and (b) spiked with of freedom equal to the number of ratio values minus 1. TSR-022 at 1350, 450, and 150 μg/mL. The drug removal treatment was applied to drug tolerance samples prior to Method validation analysis. Full method validation was completed for the aspects of intra- and inter-assay precision, sensitivity, selectiv- Robustness ity, drug tolerance, interference by TSR-022, hook effect, The robustness of the assay was determined by varying assay robustness, and stabilities. the incubation time as described in Table S1. Plates were run concurrently and analyzed on the same day by a sin- Sensitivity gle analyst. Assay sensitivity was defined as the concentration of anti- dostarlimab NAbs spiked in pNHS with dostarlimab at 250 μg/mL followed by drug removal treatment, which Stability consistently generated a signal above or equal to the To assess the stability of the NAb, samples of HPC, applicable cut point. The overall sensitivity was equal to LPC, and NC were stored in 3 different conditions: 4 h the average concentration of all plates + t × at ambient temperature, 24 h at 2–8 °C, or subject to 6 0.05, z (one side) SD. Multiple levels of NAb PCs were prepared by spiking freeze-thaw cycles. Zhang  et al. AAPS Open (2021) 7:8 Page 5 of 14 System suitability which results in the inhibition of the assay signal from The assay suitability criteria were established using the PD-1 and PD-L1 complex. In the presence of anti-dostar- data from 15 of the 19 validation runs (robustness was limab NAb, dostarlimab can be neutralized and unable to not used to establish criteria). bind with PD-1, which triggers the increase of assay sig- nal corresponding to a NAb-positive result (Fig. 1). Results In the study, samples from patients treated with A competitive ligand-binding assay was designed to dostarlimab may contain excess drug, potentially result- detect NAbs against dostarlimab. In brief, biotinylated ing in assay interference. Thus, the excess dostarlimab human PD-1 captured on a streptavidin-coated plate must be removed before the assay via the drug removal binds with SULFO-TAG–labeled human PD-L1 to gen- procedure (Fig.  2; see “Materials and methods” for the erate the ECL assay signal. Additionally, the intensity of drug removal procedure). The drug removal efficiency the ECL signal is proportional to the binding amount of was monitored through an enzyme-linked immuno- PD-1 and PD-L1 complex. When the constant amount assay (ELISA) assay with the validated range of 32.0– of dostarlimab is added, it competitively binds to PD-1, 814 ng/mL (data on file). This process confirmed that Fig. 1 NAb detection assay. A Complexes form between the fixed amount of added drug and NAb; the formation of these complexes allows SULFO-TAG–labeled PD-L1 to bind to biotinylated PD-1, which can bind to the streptavidin-ECL plate. B In the absence of NAb, the fixed amount of drug added to the solution binds to biotinylated PD-1 and prevents its interaction with PD-L1. SULFO-TAG–labeled PD-L1 bound to PD-1 is necessary to generate the ECL signal. ECL, electrochemiluminescence; NAb, neutralizing antibody; PD-1, programmed death 1; PD-L1, programmed death ligand 1 Fig. 2 Drug removal procedure. a Acidification of serum solution containing drug-NAb complexes with 1 M acetic acid separates the drug from the drug-NAb complex. b Free drug is captured by PD-1–biotin that is bound to streptavidin magnetic Dynabeads, allowing the free NAb to be detected in the NAb detection assay. NAb, neutralizing antibody; PD-1, programmed death 1 Zhang et al. AAPS Open (2021) 7:8 Page 6 of 14 S/N ratio values the processed samples had below quantification limit Figure  4 provides scatter diagrams with box plots and results in the ELISA assay, thus demonstrating that at variances of the serum S/N ratio values versus the assay least 250 μg/mL of the drug could be removed during run and analyst from 60 cancer sera. The boxplot data the drug removal process. The drug removal step allows indicated no statistically significant difference between the assay to gain drug tolerance of at least 250 μg/mL, runs and analysts. The Levene test of the S/N ratio values which matches the level required for the 3-tier bridging showed that the variances were equal among runs and ADA assay (data on file). analysts (Levene test, P = 0.1006). NAb assay cut point and CPF Sources of variation The NAb assay cut point (ACP) was defined as the level of Negative control assay response above or below which a sample is potentially Figure  3 illustrates the variation in NC ECL values positive or negative for the presence of NAbs. To evalu- listed across assay runs and plates. No NC values had a ate the ACP and CPF, 60 individual human serum samples high percent coefficient of variation (%CV > 20%), and underwent the drug removal process and were screened for no values were identified as statistical outliers, leav - ACP factor determination along with control samples over ing all 32 NC values in the final analysis. The Levene 8 runs using 1 plate per run. Samples with %CV > 20% were test for homogeneity of variances for runs and analysts excluded from the data set and not included in the ACP showed that the variances of NC ECL values among calculations. Overall, 92.5% of data points were included runs and analysts were equal (Levene test, P = 0.1073). Fig. 3 Plot of negative control ECL values by run and analyst. Box and whisker plot of 25th and 75th percentiles, median, minimum, and maximum is presented. DF Den, denominator degree of freedom; DF Num, numerator degree of freedom; ECL, electrochemiluminescence; Prob, probability S/N, signal to noise (See figure on next page.) Fig. 4 Variation estimates by runs and analysts using human serum sample S/N ratio values, excluding high CV samples, voided samples, and outliers. Box and whisker plot of 25th and 75th percentiles, median, minimum, and maximum is presented. Analyst 1 = (Q3 + 1.5 × interquartile); analyst 2 = (Q3 + 1.5 × interquartile). a Estimation of variation between runs. b Estimation of variation between analysts. c Estimation of variation between analysts and runs. CV, coefficient of variation; DF Den, denominator degree of freedom; DF Num, numerator degree of freedom; ECL, electrochemiluminescence; Prob, probability S/N, signal to noise Zhang  et al. AAPS Open (2021) 7:8 Page 7 of 14 Fig. 4 (See legend on previous page.) Zhang et al. AAPS Open (2021) 7:8 Page 8 of 14 in the calculation, and the overall mean of the rECL values ACP = mean NC plate × 1.18. (mean sample ECL value divided by mean plate NC ECL value) was determined. The distribution of the rECL values was found to be normal (the Shapiro-Wilk test, P = 0.3069). False‑negative error rate The skewness coefficient was within acceptable limits (P = An estimate of the LPC false-negative error rate was 0.1955), suggesting the parametric estimate could be used. determined based on the distribution of S/N ratio val- Figure  5 provides a scatter plot of the plate-specific ECL ues, where S was the ECL for the LPC and N was the values from the human serum samples versus the ECL val- mean of the ECL of the NC on the same plate. The ues of the NC on the corresponding plates. The linear rela - probability of an S/N ratio less than the parametric CPF tionship between sample ECL and NC ECL is shown with of 1.18 was determined by calculating the probability a slope of 0.92 (P < 0.0001), supporting the application of a of a t-distribution with 15 degrees of freedom having a floating CPF for screening test samples for the presence of t-value less than ([NAb CPF − mean of S/N ratio val- dostarlimab NAbs. ues]/SD of S/N ratio values). The false-negative error After excluding the statistical outliers by box plot analysis rate was computed to be < 0.1 % with all the LPC (500 (high limit = Q3 + 1.5 × [Q3 − Q1] and low limit = Q1 − ng/mL) S/N ratio values being above the 1.18 factor. 1.5 × [Q3 − Q1], where Q3 is the 75th percentile and Q1 is The false-negative rate is a regulatory concern. With a the 25th percentile), the ACP was calculated by parametric defined CPF, the likelihood for samples with NAb con - analysis using the following equations: centrations as low as 500 ng/mL to be determined as negative is minimal. CPF = mean S∕N(rECL) + 2.33 × SD, where 2.33 is the 99th percentile of the Assay validations normal distribution;ACP = CPF Sensitivity (Table 1) × mean NC plate. Assay sensitivity was determined to be 476.5 ng/mL using the formula The CPF was found to be 1.18; thus, the plate-specific cut Assay sensitivity = 339.3 ng∕mL + (1.645 × 83.4) = 476.5 ng∕mL. points were calculated using the following equation: Fig. 5 Evaluation of human serum ECL values by corresponding negative control, excluding high CV samples, voided samples, and outliers. Linear fit: ECL = − 147.9245 + 0.9242683 × NC. ECL, electrochemiluminescence; NC, negative control Zhang  et al. AAPS Open (2021) 7:8 Page 9 of 14 Table 1 Assay sensitivity Table 3 Inter-assay precision Run ID CP corresponding Overall inter‑assay precision %CV concentration (ng/ mL) High-positive control 19.0 Low-positive control 1 23.9 Run 4 387.9 Low-positive control 2 14.8 Run 6 366.2 Negative control 8.0 Run 9 431.0 %CV coefficient of variation Run 10 221.4 Run 11 289.8 Mean 339.3 SD 83.4 of NAb concentrations higher than 1250 ng/mL were Sensitivity 476.5 slightly decreased, a large hook effect was not observed at higher NAb concentrations. Therefore, a positive sig - CP cut point, SD standard deviation a ^ nal would still be detected if the sample contained up to Regression model: 5PL (auto) with weighting factor of 1/F 2 20,000 ng/mL anti-dostarlimab NAbs. Sensitivity was calculated using “average concentration + t 0.05, Z (one side) × SD” Selectivity Overall, 95% of the spiked samples (human cancer serum, The established LPC mean + t × SD = 339.3 ng∕mL + [2.33 × 83.4] 0.01,z (one side) hemolyzed normal serum, and lipemic normal serum) was 533.6 ng∕mL with a 1%false − negative rate. had mean ECL values greater than the plate-specific cut point, and 90% of the unspiked samples had mean ECL Assay precision values less than the plate-specific cut point (Table  4). Intra-assay precision was evaluated by running 6 inde- u Th s, the selectivity of the method met the acceptance pendent validation samples for each level (NC, LPC, and criteria with all 3 serum types: cancer serum, hemolyzed HPC) on a single plate. The %CV of the intra-assay preci - normal serum, and lipemic normal serum. sion was 7.7, 7.5, and 8.1 for the HPC, LPC (500 ng/mL), and NC respectively (Table 2). The %CV of the inter-assay Drug interference precision was 19.0, 23.9, 14.8, and 8.0 for the HPC, LPC-1 As shown in Table  5, all TSR-022–spiked NC samples (500 ng/mL), LPC-2 (533.6 ng/mL), and NC, respectively had ECL values below the cut point, and TSR-022–spiked (Table  3). Therefore, the intra-assay precision and inter- LPC samples had ECL values above the cut point. There - assay precision results met the specified criteria of %CV fore, as long as the drug removal step is performed, even ≤ 20% and %CV ≤ 25%, respectively (Gupta et al. 2011). high levels of TSR-022 (up to 1350 μg/mL) do not inter- fere with the signal for the LPC or NC. Hook effect Results in Table S2 show that the ECL signal of the NAb Assay robustness sample with the highest concentration (20,000 ng/mL) All robustness runs met the acceptance criteria was above the cut point. Even though the signal levels (Table  S3). Therefore, the assay was deemed robust with an incubation time ranging from 30 to 90 minutes per step: plate blocking, dostarlimab-NAb incubation, or sample incubation in MSD plate (Table S1). Table 2 Intra-assay precision Quality control # High‑positive Low‑positive Negative Stability control control control To assess the stability of NAbs, samples of HPC, LPC, 1 9.4 7.4 1.0 and NC were stored at 3 different conditions: 4 h at ambi - 2 9.2 7.3 0.9 ent temperature, 24 h at 2–8 °C, or subject to 6 freeze- 3 9.8 7.7 0.9 thaw cycles. All ECL signals of samples assessed for stability were within the established acceptance criteria 4 11.0 8.5 1.1 (Table  S4). Therefore, NAbs passed all the stability tests 5 9.2 7.7 1.0 performed. 6 10.4 8.7 1.1 Mean rECL 9.8 7.9 1.0 System suitability criteria %CV 7.7 7.5 8.1 The assay suitability criteria were used as part of the run %CV coefficient of variation, ECL electrochemiluminescence, rECL mean sample ECL value divided by mean plate negative control ECL value acceptance criteria to determine whether a run should Zhang et al. AAPS Open (2021) 7:8 Page 10 of 14 Table 4 Assay selectivity Sample type Lot NAb unspiked NAb spiked at LPC ECL %CV Plate CP Condition ECL %CV Plate CP Condition Cancer 1 1113 2.9 1573 < CP 2745 7.2 1573 > CP Cancer 2 1499 3.5 1573 < CP 2644 0.1 1573 > CP Cancer 3 1917 0.8 1573 > CP 2949 5.7 1573 > CP Cancer 4 1352 2.1 1573 < CP 2820 4.9 1573 > CP Cancer 5 1389 9.5 1573 < CP 2390 6.4 1573 > CP Cancer 6 1269 0.8 1573 < CP 3382 4.9 1573 > CP Cancer 7 1005 1.4 1573 < CP 2402 4.4 1573 > CP Cancer 8 998 7.4 1573 < CP 2362 1.2 1573 > CP Cancer 9 967 13.8 1573 < CP 1932 10.7 1573 > CP Cancer 10 836 11.9 1573 < CP 2478 1.3 1573 > CP Hemolyzed 1 1071 12.5 1573 < CP 2308 4.0 1573 > CP Hemolyzed 2 1144 0.8 1573 < CP 2745 19.9 1573 > CP Hemolyzed 3 893 9.0 1573 < CP 1717 3.4 1573 > CP Hemolyzed 4 802 0.4 1573 < CP 2066 10.7 1573 > CP Hemolyzed 5 553 2.0 1573 < CP 691 6.0 1573 < CP Lipemic 1 2561 6.3 1573 > CP 3486 5.7 1573 > CP Lipemic 2 1212 7.3 1573 < CP 2042 11.1 1573 > CP Lipemic 3 813 1.4 1573 < CP 1652 5.2 1573 > CP Lipemic 4 1525 6.0 1573 < CP 2758 1.0 1573 > CP Lipemic 5 1010 0.8 1573 < CP 1916 10.6 1573 > CP CP, cut point; %CV, coefficient of variation; ECL, electrochemiluminescence readout; LPC, low-positive control; NAb, neutralizing antibody Table 5 Drug interference TSR‑022 (μg/mL) NAb spiked at LPC level NAb unspiked ECL %CV Plate CP Condition ECL %CV Plate CP Condition 1350 14676 7.4 2484.5 > CP 2253 10.1 2484.5 < CP 450 11298 12.7 2484.5 > CP 2070 6.1 2484.5 < CP 150 11444 8.8 2484.5 > CP 1962 8.5 2484.5 < CP 0 13930 8.1 2484.5 > CP 2391 10.8 2484.5 < CP CP cut point, %CV coefficient of variation, ECL electrochemiluminescence readout, LPC low-positive control, NAb neutralizing antibody be rejected or accepted during validation stability testing Discussion and study sample analysis. The following system suitabil - Assay design ity criteria were obtained: Initially, a cell-based assay was chosen to determine if it would meet the acceptance criteria to function as the • NC range: rECL 0.76–1.24 NAb assay (Cong et  al. 2015). This cell-based biolumi - • LPC-1 range: rECL 1.82–11.01 nescent reporter assay was developed to quantify the • LPC-2 range: rECL 4.62–12.00 NAb to the therapeutic mAbs for either anti–PD-1 or • HPC range: rECL 5.11–18.36 anti–PD-L1 that block the PD-1/PD-L1 interaction. • NC < ACP < LPC < HPC PD-1 effector cells that express a nuclear factor of acti - vated T cells (NFAT)-luciferase reporter and a human PD-1 receptor were engineered from Jurkat T cells (Cong Analysis of all the validation runs showed that all QCs et  al. 2015). When the engineered Jurkat T cells were and NCs were within the acceptable range. Zhang  et al. AAPS Open (2021) 7:8 Page 11 of 14 cocultured with antigen-presenting cells that expressed detection. Furthermore, unlike most other cell-based T cell receptor (TCR) activator and PD-L1, activation of assays, which require only 1 type of cell line, the PD-1/ the NFAT pathway through TCR activator/TCR complex PD-L1 blockade bioassay required culturing, mainte- interaction would normally occur, leading to biolumines- nance, and preparation of 2 independent cell lines, fol- cence (Cong et  al. 2015). However, engagement of PD-1 lowed by co-culturing in an appropriate ratio to mimic and PD-L1 on the cells would block the NFAT pathway PD-1/PD-L1 function in vivo. Undoubtedly, this would and bioluminescence. Both anti–PD-1 and anti–PD-L1 potentially create more uncertainty leading to less sen- mAbs were able to block the PD-1/PD-L1 interaction, sitive and less reproducible results. resulting in the recovery of NFAT pathway signaling and The drug tolerance of this NAb assay was at least 250 bioluminescence (Cong et al. 2015). This assay, which can μg/mL with the implementation of the drug removal be used as a potency assay for anti–PD-1 drugs, was our process. This level of tolerance is based on the expo - starting point for NAb assay development. However, the sure level observed from the recommended therapeu- assay was labor intensive, had low tolerance for serum, tic dose (4 doses of 500 mg Q3W followed by 1000 mg and lacked sensitivity (data on file), which did not allow Q6W afterwards) established in the GARNET study. the cell-based reporter assay to move forward as a vali- The exposure evaluation was done at cycles 1, 4, 5, 8, dated anti-dostarlimab NAb assay. and 12, and only pre-dose samples were used for the The current trend for choosing a NAb assay format immunogenicity evaluation. During the treatment pro- includes using the therapeutic MOA as the driving selec- cess, the highest observed concentration, which was tion criterion and, in addition, relies on both assay perfor- confirmed ADA-positive through 3-tier ADA assays, mance characteristics and the risk of immunogenicity to was lower than 250 μg/mL. The drug tolerance of this the patient (Wu et al. 2016). When considering the MOA assay allows all samples to be evaluated with the low of the drug, both drug-target interaction and the struc- inconclusive rate. In the NAb assay validation, the cut tural characteristics need to be examined. Dostarlimab is point was determined using 60 individual human serum an antagonistic mAb that directly blocks the function of samples with the CPF of 1.18. The selection of the 60 the extracellular domains of PD-1, which are responsible cancer sera was balanced with respect to age, sex, and for the interaction with PD-L1 and signal transduction to cancer types. This CPF can be applied to different indi - the intra-cellular domains. Since dostarlimab blocks the cations without bias. The GARNET study includes mul - specific PD-1/PD-L1 interaction by binding to the PD-1 tiple patient populations, and in its NAb evaluation, it receptor, instead of binding at another site to exert its is unclear whether this pre-study cut point is suitable effect, a non–cell-based NAb assay is adequate for NAb for the population being studied (ClinicalTrials.gov detection (Wu et  al. 2016). Two other anti–PD-1 mAbs, 2016). Statistical determination of validation and in- nivolumab and pembrolizumab, also directly block the study immunogenicity cut points were performed by interaction of PD-1 with its ligands. Currently, non– the same statistician at Frontage Laboratory (Exton, PA) cell-based assays are used for the detection of NAbs to using the same procedures. An in-study cut point was nivolumab and pembrolizumab (van Vugt et  al. 2019; determined statistically. S/N ratio responses (also called Enrico et al. 2020). rECL) from 40 individual pre-dose baseline serum sam- ples from the GARNET study that were negative in the Assay performance and clinical relevance NAb assay were used for the cut point analysis. Out of A cell-based assay was evaluated before the white the 40 S/N ratios, no outliers were identified, and all paper on assay format assessment was published (Wu values were included in cut point determination. The et  al. 2016). Extensive optimization of the cell-based normality of the 40 S/N ratio responses was evaluated assay, including PD-L1 cell plating density, recovery using the Shapiro-Wilk test, and the test gave a P value of PD-1/PD-L1 co-culture or induction time, human = 0.0754, which is more than 0.01 (1%), indicating that serum concentration, cell culture plate type, luciferase the data set of 40 S/N ratio values was considered to be substrate volume and reaction time, and plate reader normally distributed. setting, were completed to improve the assay sensitiv- The in-study parametric CPF for the NAb was cal - ity. Under optimal assay conditions, the sensitivity in culated as 1.35 at a 1% false-positive error rate. In the the presence of 0.4 μg/mL of the drug was determined original NAb sample analysis based on the method vali- to be 12 μg/mL by screening 50 lots of human can- dation CPF of 1.18 at the 1% false-positive error rate for cer serum. Since this was far from the USA Food and the GARNET study, 64 of 135 confirmed ADA-positive Drug Administration recommended assay sensitiv- samples were determined to be NAb positive. Upon ity of 100 ng/mL, it was concluded that this assay was retrospective application of the higher in-study cut highly insensitive and was not serviceable for NAb point, 18 of the 64 samples switched from NAb positive Zhang et al. AAPS Open (2021) 7:8 Page 12 of 14 to NAb negative. Based on the 3-tier ADA assays, 13 tested and found to be acceptable. The performance of patients had treatment-emergent ADAs. Of these 13 the assay was independent of runs and analysts. With the patients, 7 patients were confirmed as NAb positive system’s suitability, stability, and drug tolerance, which and the other 6 were NAb negative using both the vali- covers the drug concentration range for the therapeu- dation cut point and the in-study cut point. Use of the tic dose established, this assay has been used to evalu- higher in-study NAb cut point did not change the num- ate NAbs to support the clinical trials that help guide the ber of patients with treatment-emergent ADAs who development of dostarlimab. were positive for NAbs, and thus would have no effect on the overall interpretation of the immunogenic- Abbreviations ity findings for this study using the tiered approach of ACP: Assay cut point; ADAs: Anti-drug antibodies; CPFs: Cut point factors; CV: 3-tier ADA assays and this NAb assay. Furthermore, Coefficient of variation; dMMR: Mismatch repair-deficient; ECL: Electrochemilu- minescence; ELISA: Enzyme-linked immunoassay; HPC: High-positive control; method sensitivity and drug tolerance should not be LPC: Low-positive control; mAb: Monoclonal antibody; MOA: Mechanism of greatly impacted by the lot of serum samples used for action; MSD: Meso Scale Diagnostics; NAbs: Neutralizing antibodies; NC: Nega- the establishment of cut point(s). The 60 lots of serum tive control; NFAT: Nuclear factor of activated T cells; PBS: Phosphate-buffered saline; PD-1: Programmed death 1; PD-L1: Programmed death ligand 1; PK: samples were selected to balance for cancer type, age, Pharmacokinetic; pNHS: Pooled normal human serum; SD: Standard deviation; and sex. In addition, only 40 pre-dose samples from S/N: Signal to noise; TCR : T cell receptor. the study were used to calculate the in-study cut point, which was not optimal. Method sensitivity and drug Supplementary Information tolerance should be evaluated by spiking in the pooled The online version contains supplementary material available at https:// doi. pre-dose samples of the study. This was not required org/ 10. 1186/ s41120- 021- 00039-w. here since the validation cut point can be further used due to limited impact as demonstrated in the above Additional file 1: Table S1. Validation of Assay’s Robustness. Table S2. Hook Eec ff t. Table S3. Assay’s Robustness. Table S4. Assay Stability analysis results. With the limited positive ADA data in the GARNET study, no clinically relevant impact was observed from PK, safety, or efficacy, which is consist - Acknowledgements Medical writing and editorial assistance, funded by GlaxoSmithKline ent with the observation that in currently approved ( Waltham, Massachusetts) and coordinated by Hasan Jamal, MSc, of Glaxo- anti–PD-1 mAbs, the presence of ADAs and NAbs SmithKline, were provided by Shannon Morgan-Pelosi, PhD, and Jennifer does not correlate with an impact on pharmacokinet- Robertson, PhD, of Ashfield MedComms, an Ashfield Health company (Mid- dletown, CT, USA). ics, pharmacodynamics, safety, or efficacy (Davda et al. 2019; Enrico et al. 2020). Authors’ contributions Allauthors provided substantial contributions to the conception or design of thework, or the acquisition, analysis, or interpretation of data for the work. Conclusions Allauthors drafted the work, revised it critically for important intellectualcon- tent, and provided final approval of the version to be published. Allauthors Because the clinical implications of ADAs and NAbs had access to full data and analyses presented in this manuscript. against dostarlimab remain unclear, both ADAs and NAbs to dostarlimab should be accurately detected Authors’ information Not applicable using an established assay, such as that described here (Enrico et  al. 2020). As has been demonstrated with Funding other anti–PD-1 mAbs, a validated non–cell-based com- Funding for this study was provided by GlaxoSmithKline (NCT02715284). Trademarks are owned by or licensed to the GSK group of companies. petitive ligand-binding assay is appropriate for detecting NAbs against dostarlimab with suitable precision, sensi- Availability of data and materials tivity, and selectivity (Enrico et al. 2020; Center for Drug GSK makes available anonymized individual participant data and associated documents from interventional clinical studies which evaluate medicines, Evaluation and Research 2017). Detection of ADAs, upon approval of proposals submitted to www. clini calst udyda tareq uest. com. especially NAbs, is extremely important for the evalua- To access data for other types of GSK sponsored research, for study docu- tion of efficacy and safety of biologic therapeutics. NAbs ments without patient-level data, and for clinical studies not listed, please submit an inquiry via the website. act by neutralizing biologic therapeutics; therefore, their presence may diminish drug efficacy. To this end, we Declarations have established an assay that can detect NAbs against dostarlimab, a humanized anti–PD-1 antibody. We have Competing interests determined the CPF for the detection of neutralizing Xiaolong Tom Zhang has nothing to disclose. Hong Chen has nothing to dis- close. Weiping Shao has nothing to disclose. Zhongping John Lin has nothing anti-dostarlimab antibodies in human serum at a 1% to disclose. Murad Melhem is an employee of GlaxoSmithKline. Sharon Lu is an false-positive rate. The assay’s precision, sensitivity, hook employee of Scholar Rock and was affiliated with GlaxoSmithKline at the time effect, selectivity, robustness, and drug interference were of the study analyses. Zhang  et al. AAPS Open (2021) 7:8 Page 13 of 14 Author details review of registration trials and future considerations. J Immunother WuXi AppTec, 288 Fute Zhong Road, Waigaoqiao Free Trade Zone, Shang- Cancer 6(1):8 hai 200131, China. Alliance Pharma, 17 Lee Boulevard, Malvern, PA 19355, Gunn GR, Sealey DCF, Jamali F, Meibohm B, Ghosh S, Shankar G (2016) From USA. Frontage Laboratories, 700 Pennsylvania Drive, Exton, PA 19341, USA. the bench to clinical practice: understanding the challenges and GlaxoSmithKline, 1000 Winter Street, Waltham, MA 02451, USA. uncertainties in immunogenicity testing for biopharmaceuticals. Clin Exp Immunol 184(2):137–146 Received: 4 June 2021 Accepted: 13 October 2021 Gupta S, Devanarayan V, Finco D, Gunn GR III, Kirshner S, Richards S et al (2011) Recommendations for the validation of cell-based assays used for the detection of neutralizing antibody immune responses elicited against biological therapeutics. J Pharm Biomed Anal 55(5):878–888 Jolicoeur P, Tacey RL (2012) Development and validation of cell-based assays for the detection of neutralizing antibodies to drug products: a practical References approach. Bioanalysis. 4(24):2959–2970 Ahn BC, Pyo KH, Xin CF, Jung D, Shim HS, Lee CY et al (2019) Comprehensive Kaufman HL, Russell J, Hamid O, Bhatia S, Terheyden P, D’Angelo SP et al (2016) analysis of the characteristics and treatment outcomes of patients with Avelumab in patients with chemotherapy-refractory metastatic Merkel non-small cell lung cancer treated with anti-PD-1 therapy in real-world cell carcinoma: a multicentre, single-group, open-label, phase 2 trial. practice. J Cancer Res Clin Oncol 145(6):1613–1623 Lancet Oncol 17(10):1374–1385 Andre T, Berton D, Curigliano G, Ellard S, Perez JMT, Arkenau HT et al (2021) Khozin S, Miksad RA, Adami J, Boyd M, Brown NR, Gossai A et al (2019) Real- Safety and efficacy of anti-PD-1 antibody dostarlimab in patients with world progression, treatment, and survival outcomes during rapid adop- mismatch repair-deficient (dMMR) solid cancer: results from the GARNET tion of immunotherapy for advanced non-small cell lung cancer. Cancer. study. J Clin Oncol 39(suppl 3):abstr 9 125(22):4019–4032 Ansell SM, Lesokhin AM, Borrello I, Halwani A, Scott EC, Gutierrez M et al Krishna M, Nadler S (2016) Immunogenicity to biotherapeutics - the role of (2015) PD-1 blockade with nivolumab in relapsed or refractory Hodgkin’s anti-drug immune complexes. Front Immunol 7:21 lymphoma. N Engl J Med 37(4):311–319 Levene H (1960) Robust tests for the equality of variances. In: Oaklin I (ed) Antonia SJ, Villegas A, Daniel D, Vicente D, Murakami S, Hui R, PACIFIC Inves- Contributions to probability and statistics. Stanford University Press, Palo tigators et al (2017) Durvalumab after chemoradiotherapy in stage III Alto, pp 278–292 non-small-cell lung cancer. N Engl J Med 377(20):1919–1929 Liao K, Sikkema D, Wang C, Chen K, DeWall S, Lee TN (2012) Development and Apolo AB, Ellerton JA, Infante JR, Agrawal M, Gordon MS, Aljumaily R et al validation of a cell-based SEAP reporter assay for the detection of neutral- (2017) Updated efficacy and safety of avelumab in metastatic urothelial izing antibodies against an anti-IL-13 therapeutic antibody. J Immunol carcinoma (mUC): pooled analysis from 2 cohorts of the phase 1b Javelin Methods 375(1–2):258–263 solid tumor study. J Clin Oncol 35(15 suppl):4528 Mire-Sluis AR, Barrett YC, Devanarayan V, Koren E, Liu H, Maia M et al (2004) Bloem K, Hernández-Breijo B, Martínez-Feito A, Rispens T (2017) Immuno- Recommendations for the design and optimization of immunoassays genicity of therapeutic antibodies: monitoring antidrug antibodies in a used in the detection of host antibodies against biotechnology products. clinical context. Ther Drug Monit 39(4):327–332 J Immunol Methods 289:1–16 Center for Drug Evaluation and Research (2017) Approval package for: Oaknin A, Ellard SL, Leath C III, Moreno V, Kristeleit R, Guo W et al (2018) application number 125514Orig1s014. https:// www. acces sdata. fda. Preliminary safety, efficacy, and PK/PD characterization from GARNET, a gov/ drugs atfda_ docs/ nda/ 2019/ 12551 4Orig 1s014. pdf. Accessed 4 phase 1 clinical trial of the anti–PD-1 monoclonal antibody, TSR-042, in Feb 2021. patients (pts) with recurrent or advanced MSI-H endometrial cancer. Ann ClinicalTrials.gov (2016) Study of TSR-042, an anti-programmed cell death-1 Oncol 29(suppl8):viii334 receptor (PD-1) monoclonal antibody, in participants with advanced Pineda C, Hernandez GC, Jacobs IA, Alvarez DF, Carini C (2016) Assessing the solid tumors (GARNET ): NCT02715284. https:// clini caltr ials. gov/ ct2/ show/ immunogenicity of biopharmaceuticals. BioDrugs. 30:195–206 NCT02 715284. Accessed 6 May 2020. https:// clini caltr ials. gov/ ct2/ show/ Robert C, Ribas A, Wolchok JD, Hodi FS, Hamid O, Kefford R et al (2014) NCT02 715284? term= 02715 284& draw= 2& rank=1 Anti-programmed-death-receptor-1 treatment with pembrolizumab in Cong M, Cheng Z, Karassina N, Grailer J, Hartnett J, Cosby N, Fan F et al (2015) ipilimumab-refractory advanced melanoma: a randomised dose-compar- Advertorial: novel bioassay to assess PD-1/PD-L1 therapeutic antibodies ison cohort of a phase 1 trial. Lancet. 384(9948):1109–1117 in development for immunotherapy. Genet Eng News 35(10) https:// Seiwert TY, Burtness B, Mehra R, Weiss J, Berger R, Eder JP et al (2016) Safety www. genen gnews. com/ magaz ine/ 247/ adver torial- novel- bioas say- to- and clinical activity of pembrolizumab for treatment of recurrent or assess- pd-1- pd- l1- thera peutic- antib odies- in- devel opment- for- immun metastatic squamous cell carcinoma of the head and neck (KEY- other apy/. Accessed 14 Aug 2020 NOTE-012): an open-label, multicentre, phase 1b trial. Lancet Oncol Davda J, Declerck P, Hu-Lieskovan S, Hickling TP, Jacobs IA, Chou J et al (2019) 17(7):956–965 Immunogenicity of immunomodulatory, antibody-based, oncology Shankar G, Devanarayan V, Amaravadi L, Barrett YC, Bowsher R, Finco-Kent D therapeutics. J Immunother Cancer 7:105 et al (2008) Recommendations for the validation of immunoassays used El-Khoueiry AB, Sangro B, Yau T, Crocenzi TS, Kudo M, Hsu C et al (2017) for detection of host antibodies against biotechnology products. J Pharm Nivolumab in patients with advanced hepatocellular carcinoma (Check- Biomed Anal 48(5):1267–1281 Mate 040): an open-label, non-comparative, phase 1/2 dose escalation Shapiro SS, Wilk MB (1965) Analysis of variance test for normality (complete and expansion trial. Lancet. 389(10088):2492–2502 samples). Biometrika. 52:591–611 Enrico D, Paci A, Chaput N, Karamouza E, Besse B (2020) Antidrug antibodies US Food and Drug Administration. (2021a) FDA grants accelerated approval against immune checkpoint blockers: impairment of drug efficacy or to dostarlimab-gxly for dMMR endometrial cancer. https:// www. fda. indication of immune activation? Clin Cancer Res 26(4):787–792 gov/ drugs/ drug- appro vals- and- datab ases/ fda- grants- accel erated- European Medicines Agency (2021) Jemperli. https:// www. ema. europa. eu/ en/ appro val- dosta rlimab- gxly- dmmr- endom etrial- cancer. Accessed 24 medic ines/ human/ EPAR/ jempe rli. Accessed 24 May 2020 May 2021. Fehrenbacher L, Spira A, Ballinger M, Kowanetz M, Vansteenkiste J, Mazieres US Food and Drug Administration. (2021b) FDA grants accelerated approval to J, POPLAR Study Group et al (2016) Atezolizumab versus docetaxel for dostarlimab-gxly for dMMR advanced solid tumors. https:// www. fda. gov/ patients with previously treated non-small-cell lung cancer (POPLAR): drugs/ resou rces- infor mation- appro ved- drugs/ fda- grants- accel erated- a multicentre, open-label, phase 2 randomised controlled trial. Lancet. appro val- dosta rlimab- gxly- dmmr- advan ced- solid- tumors. Accessed 16 387(10030):1837–1846 Sept 2021. Garcês S, Demengeot J (2018) The immunogenicity of biologic therapies. Curr van Vugt MJH, Stone JA, De Greef RHJMM, Snyder ES, Lipka L, Turner DC et al Probl Dermatol 53:37–48 (2019) Immunogenicity of pembrolizumab in patients with advanced Gong J, Chehrazi-Raffle A, Reddi S, Salgia R (2018) Development of PD-1 and tumors. J Immunother Cancer 7(1):212 PD-L1 inhibitors as a form of cancer immunotherapy: a comprehensive Zhang et al. AAPS Open (2021) 7:8 Page 14 of 14 Weber JS, D’Angelo SP, Minor D, Hodi FS, Gutzmer R, Neyns B et al (2015) Wu Y, Akhgar A, Li J, Yu B, Chen C, Lee N et al (2018) Selection of a ligand- Nivolumab versus chemotherapy in patients with advanced mela- binding neutralizing antibody assay for benralizumab: comparison with noma who progressed after anti-CTLA-4 treatment (CheckMate 037): an antibody-dependent cell-mediated cytotoxicity (ADCC) cell-based a randomised, controlled, open-label, phase 3 trial. Lancet Oncol assay. AAPS. 20(3):49 16(4):375–384 Wu B, Chung S, Jiang XR, McNally J, Pedras-Vasconcelos J, Pillutla R et al (2016) Publisher’s Note Strategies to determine assay format for the assessment of neutralizing Springer Nature remains neutral with regard to jurisdictional claims in pub- antibody responses to biotherapeutics. AAPS J 18(6):1335–1350 lished maps and institutional affiliations. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png AAPS Open Springer Journals

A competitive ligand-binding assay for the detection of neutralizing antibodies against dostarlimab (TSR-042)

Download PDF
14 pages

Loading next page...
 
/lp/springer-journals/a-competitive-ligand-binding-assay-for-the-detection-of-neutralizing-MJrKUKtsxU

References (31)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2021
eISSN
2364-9534
DOI
10.1186/s41120-021-00039-w
Publisher site
See Article on Publisher Site

Abstract

Dostarlimab is a humanized anti–PD-1 monoclonal antibody. Dostarlimab (JEMPERLI; TSR-042) was recently approved in the USA and in the EU. The presence of neutralizing antibodies (NAbs) is a cause for concern because they block the therapeutic function of the antibody and reduce drug efficacy. Therefore, programs developing therapeutic biologics need to develop and validate assays that adequately assess the presence of NAbs in the serum of patients treated with biologic therapies. Presented here is the development and validation of a competitive ligand-binding assay that specifically detects anti-dostarlimab NAbs in human serum. Precision, sensitivity, hook effect, selectivity, assay robustness, stabilities, and system suitability were evaluated. In addition, drug tolerance of the assay with the implementation of a drug removal process was investigated. The cut point factor for the detection of NAbs in human serum at a 1% false-positive rate was determined. The assay’s precision, sensitivity, hook effect, selectivity, robustness, and drug interference were tested and found to be acceptable. With system suitability and stability established, this assay has been used to evaluate NAbs to guide the development of dostarlimab. Trial registration: Clinicaltrials.gov, NCT02 715284. Registered 9 March 2016 Keywords: Dostarlimab, Endometrial cancer, Neutralizing antibodies, Immunotherapy, Competitive ligand-binding assay Introduction monoclonal antibody (mAb) (Oaknin et  al. 2018). In the Immunotherapy with antibodies directed at the pro- USA, dostarlimab was recently approved as a mono- grammed death 1 (PD-1)/PD ligand 1(PD-L1) axis has therapy in adult patients with mismatch repair-deficient shown remarkable activity in a variety of tumor types (dMMR) recurrent or advanced endometrial cancer that in clinical trials and has been approved worldwide for has progressed on or after a platinum-containing regi- a number of malignancies (Robert et  al. 2014; Weber men or dMMR recurrent or advanced solid tumors that et al. 2015; Fehrenbacher et al. 2016; Antonia et al. 2017; have progressed on or following prior treatment and Apolo et  al. 2017; Seiwert et  al. 2016; Ansell et  al. 2015; who have no satisfactory alternative treatment options El-Khoueiry et  al. 2017; Kaufman et  al. 2016; Gong (US Food and Drug Administration 2021a; US Food et  al. 2018; Ahn et  al. 2019; Khozin et  al. 2019). Dostar- and Drug Administration 2021b; Andre et  al. 2021; limab (JEMPERLI; TSR-042) is a humanized anti–PD-1 ClinicalTrials.gov 2016). In the EU, dostarlimab was recently approved as a monotherapy in adult patients with recurrent or advanced dMMR/microsatellite insta- bility–high endometrial cancer that has progressed on *Correspondence: slu@scholarrock.com GlaxoSmithKline, 1000 Winter Street, Waltham, MA 02451, USA or after treatment with a platinum-containing regimen Full list of author information is available at the end of the article (Oaknin et al. 2018; Andre et al. 2021; ClinicalTrials.gov Weiping Shao was employed by Frontage Laboratories when the 2016; European Medicines Agency 2021). As with other research was conducted. © The Author(s) 2021. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. Zhang et al. AAPS Open (2021) 7:8 Page 2 of 14 immunotherapeutic agents, anti-drug antibodies (ADAs) Medicines Agency recognize both cell-based and non– may emerge and be detectable after administration of cell-based competitive ligand-binding assays as valid dostarlimab (Pineda et al. 2016). However, dostarlimab is measurements of NAbs in some cases (Jolicoeur and a humanized mAb, and as such, has a lower risk of elic- Tacey 2012; Wu et al. 2016). The selection of either cell- iting ADAs or neutralizing (NAbs) than non-humanized based or non–cell-based assays is driven by several dif- mAbs (Davda et al. 2019). ferent variables, including therapeutic mechanism of However, the production of ADAs is generally a treat- action (MOA), assay performance characteristics, and ment risk with therapeutic antibodies and can lead to an risk of immunogenicity, with therapeutic MOA being altered pharmacokinetic (PK) profile, which in turn may the primary determinant (Wu et al. 2016). impact the safety and efficacy of therapeutic antibodies Here, we present the development and validation of a (Garcês and Demengeot 2018; Bloem et al. 2017). ADAs competitive ligand-binding assay that is specific for the can be (1) binding, leading to minimal or no impact; (2) detection of anti-dostarlimab NAbs in human serum. We clearing, leading to impact through an altered PK pro- validated the assay’s precision, sensitivity, hook effect, file; (3) sustaining, leading to a prolonged exposure that selectivity, robustness, stability, and system suitability. In may change the efficacy and/or safety; or (4) neutralizing, addition, we investigated the drug tolerance of the assay leading to reduced pharmacological activity of the thera- with the implementation of a drug removal process, as peutic antibody (Garcês and Demengeot 2018; Bloem well as the possibilities for false-negative results. The util - et al. 2017). ity of this assay as it pertains to the dostarlimab clinical u Th s, the presence of NAbs is concerning because they development program is also briefly discussed. block the therapeutic function of the antibody and are likely to lead to reduced efficacy (Garcês and Demengeot Materials and methods 2018; Bloem et  al. 2017; Gunn et  al. 2016). Further- Materials more, NAbs may lead to safety issues, such as hypersen- Critical reagents for this assay were as follows: a mouse sitivity and anaphylaxis Gunn et  al. 2016; Krishna and anti-dostarlimab mAb (Precision Antibodies, Columbia, Nadler 2016). Therefore, programs developing therapeu - MD) generated from clone #6G10 at a concentration of tic antibodies need to develop and validate assays that 1.15 mg/mL; dostarlimab (WuXi, Philadelphia, PA) at a adequately assess the presence of NAbs in the serum of concentration of 20.7 mg/mL; biotinylated human PD-1 patients treated with such therapeutic agents. (ACRO Biosystems, Newark, DE) at a concentration of Cell-based reporter assays have been developed and 200 μg/mL; and SULFO-TAG–labeled human PD-L1 validated for detecting NAbs, including a cell-based bio- (CR0; Meso Scale Diagnostics [MSD], Rockville, MD) at luminescent reporter assay that quantifies therapeutic a concentration of 2106 μg/mL. mAbs against PD-1 and PD-L1 (Liao et  al. 2012; Cong et al. 2015; Jolicoeur and Tacey 2012). Preparation of dostarlimab drug solution Regulatory agencies prefer cell-based assays because The initial stock solution of dostarlimab was diluted they rely on cellular responses to NAb-mediated inhi- in LowCross-Buffer (Boca Scientific, Westwood, MA) bition of therapeutic antibodies (Wu et  al. 2016). to a concentration of 1608 ng/mL, which is 6 times the Because of their mode of detection, these assays have desired final concentration (268 ng/mL). The prepared been considered more biologically relevant than non– dostarlimab solution was aliquoted and stored at − 70 °C. cell-based assays (Bloem et  al. 2017; Liao et  al. 2012; Cong et  al. 2015; Jolicoeur and Tacey 2012). However, cell-based assays are labor intensive and highly vari- Preparation of assay control samples able and exhibit low serum tolerance and poor drug The negative control (NC) was prepared by adding 250 tolerance (Bloem et  al. 2017; Liao et  al. 2012; Cong μg/mL dostarlimab (GSK, Waltham, MA) into pooled et  al. 2015; Jolicoeur and Tacey 2012; Wu et  al. 2016). normal human serum (pNHS; Bio IVT, Westbury, NY) In contrast to cell-based assays, non–cell-based assays and freezing the solution for 12 h. After thawing, dostar- often rely on binding of the drug and target for signal limab was removed from the NC samples via the drug detection and quantitation. Competitive ligand-binding removal procedure (see below). The low-positive control assays, used to characterize NAbs, have been shown (LPC) and high-positive control (HPC) were prepared to provide higher sensitivity, a wider dynamic range, by mixing dostarlimab and anti-dostarlimab NAb clone increased precision, and better matrix tolerance than 6G10 in pNHS at 250 μg/mL and 500 ng/mL, respectively, cell-based assays (Wu et  al. 2018). Although regulatory for the LPC and at 250 μg/mL and 2000 ng/mL, respec- agencies generally prefer cell-based assays, the United tively, for the HPC. The solution was then subjected to the States Food and Drug Administration and the European same freeze–thaw–drug removal procedure as the NC. Zhang  et al. AAPS Open (2021) 7:8 Page 3 of 14 Once prepared, the NC, LPC, and HPC were stored at − dark. The final drug concentration was 268 ng/mL, which 70 °C. was selected based on the following: the linearity range for dostarlimab drug concentration to ECL signal was Drug removal procedure found to be 0–300 ng/mL; the highest drug concentra- Excess dostarlimab was removed from serum and con- tion in this range led to the lowest ECL signal in the test trol samples prior to performing the NAb detection sample; and any ECL signal increase due to NAb interac- assay. Biotinylated human PD-1 stock solution (200 μg/ tion in the assay introduced a better signal to noise ratio. mL) was prepared by dissolving 200 μg of PD-1 in 1 mL Therefore, approximately 90% of the 300 ng/mL (which of dH O. Magnetic streptavidin Dynabeads (Thermo was considered to be the high limit of quantification) Fisher Scientific, Waltham, MA) were prepared by trans - was used to maintain the assay reproducibility. From ferring 140 μL of streptavidin magnetic beads to a tube each well of the dilution plate, 50 μL was transferred to (approximately 50% of bead volume) where they were the washed streptavidin-coated plate and incubated for subsequently washed once using the magnetic stand with 1 hour at ambient temperature in the dark. Following 0.5 mL of wash/bind buffer. To bind human PD-1 to the incubation, the wells were washed with Tween-20 PBS, as magnetic beads, 70 μL (14 μg) of biotinylated PD-1 pro- described above, then 100 μL of 2× Read Buffer (MSD) tein was added to the washed beads, and the mixture was was added to each well. The plate was subsequently read incubated for 1 hour at ambient temperature while shak- on an MSD SECTOR Imager (MSD) to attain the electro- ing at 1000 rpm. The PD-1 solution was then removed chemiluminescence (ECL) signal. from the beads using the magnetic stand, and the beads were washed 3 times using the magnetic stand with 0.5 Statistical methods mL wash/bind buffer. Data description The positive control (PC) was prepared by adding Sixty individual human cancer serum samples were dostarlimab and NAb to pNHS to achieve a final concen - obtained from the disease population, were divided into 4 tration of 250 μg/mL of dostarlimab and the desired con- groups with 30, 30, 40, and 20 samples in groups A, B, C, centration of NAb in 70 μL of pNHS. Drug removal was and D, respectively. The plate for group D also included accomplished by adding a one-tenth volume of 1 M ace- 11 sensitivity samples. Each group was tested twice by 2 tic acid to serum or control solutions and incubating at analysts across 3 days. Each analyst tested at least 1 set of ambient temperature for 2 h, resulting in a pH of 3.0 with sensitivity samples. The reported value for each test sam - no change of ADA affinity observed. A one-tenth volume ple was the mean ECL response from duplicate wells on of 1 M Tris base was added to neutralize the samples. a plate. The NCs were tested 4 times on each plate, and The sample solution was immediately added to the pre- the means of the duplicate wells were reported for a total washed PD-1–coated magnetic beads and incubated for 1 of 32 ECL values. The LPC and HPC samples were tested hour at ambient temperature while shaking at 1000 rpm. twice on each plate for a total of 16 mean ECL values for The samples were then separated from the streptavidin each PC. beads by magnetic separation and subsequently used in the NAb assay, or alternatively, stored at − 70 °C. Sources of variation All statistical analyses were completed using JMP Pro NAb detection assay version 12.1.0 (JMP; Cary, NC), and methods used were The streptavidin-coated Gold SECTOR plates (MSD) consistent with recommendations by Shankar et  al. were blocked with 1% bovine serum albumin (Gemini (2008). Statistical outliers were determined using a quar- Bio, West Sacramento, CA) in phosphate-buffered saline tile range outlier method. The statistical outliers in the (PBS) for 1 hour at ambient temperature with shaking upper and lower quartiles were defined as Q3 + 1.5 × at 450 rpm and thoroughly washed (3 cycles on a plate (Q3 − Q1) and Q1 − 1.5 × (Q3 − Q1), where Q3 is the washer with 0.05% Tween-20 in PBS). In a dilution plate, 75th percentile and Q1 is the 25th percentile. Outliers 20 μl of prepared dostarlimab solution (1608 ng/mL) was identified by these criteria were removed, and the model added to 60 μL of drug removal samples or controls, and was re-fit. Normality of residual values was evaluated by the resulting solutions were incubated for 1 hour at ambi- the Shapiro-Wilk test (Mire-Sluis et  al. 2004). In addi- ent temperature with shaking at 450 rpm. A Master Mix tion, the skewness coefficient was calculated as a relative solution of 12 nM biotinylated PD-1 and 12 nM SULFO- measure of symmetry (Shapiro and Wilk 1965). TAG–labeled human PD-L1 was prepared in Low Cross- If skewness coefficient was less than − 1 or greater than Buffer. Then, 40 μL of Master Mix was added to each + 1, the distribution of the data was highly skewed; if well of the dilution plate, which was incubated for 1 hour skewness was between − 1 and − 1/2 or between + 1/2 at ambient temperature with shaking at 450 rpm in the and + 1, the distribution was moderately skewed; and if Zhang et al. AAPS Open (2021) 7:8 Page 4 of 14 desired NAb concentrations that span the cut point into skewness was between − 1/2 and + 1/2, the distribution pNHS with dostarlimab at 250 μg/mL. The sensitivity was approximately symmetric. The parametric cut point samples underwent the drug removal treatment prior to estimates are typically recommended if normality dis- analysis. Six runs were performed by 2 analysts over 3 tribution is confirmed (if P > 0.05); however, if normal - separate days. ity cannot be assumed, the parametric cut point estimate may be used if the data are not highly skewed (if skew- ness is between − 1 and + 1). Assay precision Intra-assay precision was evaluated by running 6 inde- NAb assay cut point pendent validation samples for NCs, LPCs (533.6 ng/ Signal to noise (S/N) ratio values were obtained by divid- mL), and HPCs (2000 ng/mL) on a single plate. The inter- ing the ECL signal value from each individual serum assay precision was measured using the PCs and NCs in sample by the mean of the NC ECL values from the cor- all accepted validation runs. responding plate. After exclusion of analytic outliers from the set of NC values, the remaining S/N ratio values were Hook effect used for statistical analyses. Homogeneity of inter-plate To determine if the assay demonstrated a hook effect, sample variances across the 8 plates was investigated per an ultrahigh concentration (20,000 ng/mL) of NAb was Levene’s test using S/N ratio values (Levene 1960). prepared in pNHS. Serial 2-fold dilutions of this sample in pNHS generated samples at 10,000, 5000, 2500, 1250, Cut point factor determination 625, 312.5, 156.25, 78.13, and 39.06 ng/mL, which were Floating cut point factors (CPFs) for parametric and evaluated as replicates. non-parametric screening were determined using the S/N ratio values. A parametric method was used to cal- Selectivity culate robust estimates of the mean and standard devia- Ten individual cancer serum samples, 5 2% hemolyzed tion (SD) of all S/N ratio values. The parametric 1% error normal serum samples and 5 lipemic normal serum sam- rate floating CPF was then determined by multiplying the ples (450 μg/dL) containing dostarlimab at 250 μg/mL SD value by the 99th quartile of the t-distribution (with (a) unspiked and (b) spiked with anti-dostarlimab NAb degrees of freedom equal to the number of S/N ratio val- at the LPC level (533.6 ng/mL), underwent drug removal ues minus 1) and adding the product to the mean of S/N treatment and were tested. ratio values. The non-parametric 1% error rate CPF was determined by calculating the empirical 99th percentile Drug interference of S/N ratio values. Interference in the detection of LPC was evaluated by using the isotype IgG4 antibody TSR-022, which is an False‑negative error rate anti-TIM3 antibody. The results were considered accept - The false-negative error rate, defined as the probability of able if the TSR-022–spiked LPC sample values were observing an S/N ratio less than the CPF in a known pos- within their respective ranges, which were established itive sample, was calculated based on the mean and SD of during this validation. Drug interference was tested using all LPC S/N ratio values and a t-distribution with degrees 3 sets of LPC and NC (a) unspiked and (b) spiked with of freedom equal to the number of ratio values minus 1. TSR-022 at 1350, 450, and 150 μg/mL. The drug removal treatment was applied to drug tolerance samples prior to Method validation analysis. Full method validation was completed for the aspects of intra- and inter-assay precision, sensitivity, selectiv- Robustness ity, drug tolerance, interference by TSR-022, hook effect, The robustness of the assay was determined by varying assay robustness, and stabilities. the incubation time as described in Table S1. Plates were run concurrently and analyzed on the same day by a sin- Sensitivity gle analyst. Assay sensitivity was defined as the concentration of anti- dostarlimab NAbs spiked in pNHS with dostarlimab at 250 μg/mL followed by drug removal treatment, which Stability consistently generated a signal above or equal to the To assess the stability of the NAb, samples of HPC, applicable cut point. The overall sensitivity was equal to LPC, and NC were stored in 3 different conditions: 4 h the average concentration of all plates + t × at ambient temperature, 24 h at 2–8 °C, or subject to 6 0.05, z (one side) SD. Multiple levels of NAb PCs were prepared by spiking freeze-thaw cycles. Zhang  et al. AAPS Open (2021) 7:8 Page 5 of 14 System suitability which results in the inhibition of the assay signal from The assay suitability criteria were established using the PD-1 and PD-L1 complex. In the presence of anti-dostar- data from 15 of the 19 validation runs (robustness was limab NAb, dostarlimab can be neutralized and unable to not used to establish criteria). bind with PD-1, which triggers the increase of assay sig- nal corresponding to a NAb-positive result (Fig. 1). Results In the study, samples from patients treated with A competitive ligand-binding assay was designed to dostarlimab may contain excess drug, potentially result- detect NAbs against dostarlimab. In brief, biotinylated ing in assay interference. Thus, the excess dostarlimab human PD-1 captured on a streptavidin-coated plate must be removed before the assay via the drug removal binds with SULFO-TAG–labeled human PD-L1 to gen- procedure (Fig.  2; see “Materials and methods” for the erate the ECL assay signal. Additionally, the intensity of drug removal procedure). The drug removal efficiency the ECL signal is proportional to the binding amount of was monitored through an enzyme-linked immuno- PD-1 and PD-L1 complex. When the constant amount assay (ELISA) assay with the validated range of 32.0– of dostarlimab is added, it competitively binds to PD-1, 814 ng/mL (data on file). This process confirmed that Fig. 1 NAb detection assay. A Complexes form between the fixed amount of added drug and NAb; the formation of these complexes allows SULFO-TAG–labeled PD-L1 to bind to biotinylated PD-1, which can bind to the streptavidin-ECL plate. B In the absence of NAb, the fixed amount of drug added to the solution binds to biotinylated PD-1 and prevents its interaction with PD-L1. SULFO-TAG–labeled PD-L1 bound to PD-1 is necessary to generate the ECL signal. ECL, electrochemiluminescence; NAb, neutralizing antibody; PD-1, programmed death 1; PD-L1, programmed death ligand 1 Fig. 2 Drug removal procedure. a Acidification of serum solution containing drug-NAb complexes with 1 M acetic acid separates the drug from the drug-NAb complex. b Free drug is captured by PD-1–biotin that is bound to streptavidin magnetic Dynabeads, allowing the free NAb to be detected in the NAb detection assay. NAb, neutralizing antibody; PD-1, programmed death 1 Zhang et al. AAPS Open (2021) 7:8 Page 6 of 14 S/N ratio values the processed samples had below quantification limit Figure  4 provides scatter diagrams with box plots and results in the ELISA assay, thus demonstrating that at variances of the serum S/N ratio values versus the assay least 250 μg/mL of the drug could be removed during run and analyst from 60 cancer sera. The boxplot data the drug removal process. The drug removal step allows indicated no statistically significant difference between the assay to gain drug tolerance of at least 250 μg/mL, runs and analysts. The Levene test of the S/N ratio values which matches the level required for the 3-tier bridging showed that the variances were equal among runs and ADA assay (data on file). analysts (Levene test, P = 0.1006). NAb assay cut point and CPF Sources of variation The NAb assay cut point (ACP) was defined as the level of Negative control assay response above or below which a sample is potentially Figure  3 illustrates the variation in NC ECL values positive or negative for the presence of NAbs. To evalu- listed across assay runs and plates. No NC values had a ate the ACP and CPF, 60 individual human serum samples high percent coefficient of variation (%CV > 20%), and underwent the drug removal process and were screened for no values were identified as statistical outliers, leav - ACP factor determination along with control samples over ing all 32 NC values in the final analysis. The Levene 8 runs using 1 plate per run. Samples with %CV > 20% were test for homogeneity of variances for runs and analysts excluded from the data set and not included in the ACP showed that the variances of NC ECL values among calculations. Overall, 92.5% of data points were included runs and analysts were equal (Levene test, P = 0.1073). Fig. 3 Plot of negative control ECL values by run and analyst. Box and whisker plot of 25th and 75th percentiles, median, minimum, and maximum is presented. DF Den, denominator degree of freedom; DF Num, numerator degree of freedom; ECL, electrochemiluminescence; Prob, probability S/N, signal to noise (See figure on next page.) Fig. 4 Variation estimates by runs and analysts using human serum sample S/N ratio values, excluding high CV samples, voided samples, and outliers. Box and whisker plot of 25th and 75th percentiles, median, minimum, and maximum is presented. Analyst 1 = (Q3 + 1.5 × interquartile); analyst 2 = (Q3 + 1.5 × interquartile). a Estimation of variation between runs. b Estimation of variation between analysts. c Estimation of variation between analysts and runs. CV, coefficient of variation; DF Den, denominator degree of freedom; DF Num, numerator degree of freedom; ECL, electrochemiluminescence; Prob, probability S/N, signal to noise Zhang  et al. AAPS Open (2021) 7:8 Page 7 of 14 Fig. 4 (See legend on previous page.) Zhang et al. AAPS Open (2021) 7:8 Page 8 of 14 in the calculation, and the overall mean of the rECL values ACP = mean NC plate × 1.18. (mean sample ECL value divided by mean plate NC ECL value) was determined. The distribution of the rECL values was found to be normal (the Shapiro-Wilk test, P = 0.3069). False‑negative error rate The skewness coefficient was within acceptable limits (P = An estimate of the LPC false-negative error rate was 0.1955), suggesting the parametric estimate could be used. determined based on the distribution of S/N ratio val- Figure  5 provides a scatter plot of the plate-specific ECL ues, where S was the ECL for the LPC and N was the values from the human serum samples versus the ECL val- mean of the ECL of the NC on the same plate. The ues of the NC on the corresponding plates. The linear rela - probability of an S/N ratio less than the parametric CPF tionship between sample ECL and NC ECL is shown with of 1.18 was determined by calculating the probability a slope of 0.92 (P < 0.0001), supporting the application of a of a t-distribution with 15 degrees of freedom having a floating CPF for screening test samples for the presence of t-value less than ([NAb CPF − mean of S/N ratio val- dostarlimab NAbs. ues]/SD of S/N ratio values). The false-negative error After excluding the statistical outliers by box plot analysis rate was computed to be < 0.1 % with all the LPC (500 (high limit = Q3 + 1.5 × [Q3 − Q1] and low limit = Q1 − ng/mL) S/N ratio values being above the 1.18 factor. 1.5 × [Q3 − Q1], where Q3 is the 75th percentile and Q1 is The false-negative rate is a regulatory concern. With a the 25th percentile), the ACP was calculated by parametric defined CPF, the likelihood for samples with NAb con - analysis using the following equations: centrations as low as 500 ng/mL to be determined as negative is minimal. CPF = mean S∕N(rECL) + 2.33 × SD, where 2.33 is the 99th percentile of the Assay validations normal distribution;ACP = CPF Sensitivity (Table 1) × mean NC plate. Assay sensitivity was determined to be 476.5 ng/mL using the formula The CPF was found to be 1.18; thus, the plate-specific cut Assay sensitivity = 339.3 ng∕mL + (1.645 × 83.4) = 476.5 ng∕mL. points were calculated using the following equation: Fig. 5 Evaluation of human serum ECL values by corresponding negative control, excluding high CV samples, voided samples, and outliers. Linear fit: ECL = − 147.9245 + 0.9242683 × NC. ECL, electrochemiluminescence; NC, negative control Zhang  et al. AAPS Open (2021) 7:8 Page 9 of 14 Table 1 Assay sensitivity Table 3 Inter-assay precision Run ID CP corresponding Overall inter‑assay precision %CV concentration (ng/ mL) High-positive control 19.0 Low-positive control 1 23.9 Run 4 387.9 Low-positive control 2 14.8 Run 6 366.2 Negative control 8.0 Run 9 431.0 %CV coefficient of variation Run 10 221.4 Run 11 289.8 Mean 339.3 SD 83.4 of NAb concentrations higher than 1250 ng/mL were Sensitivity 476.5 slightly decreased, a large hook effect was not observed at higher NAb concentrations. Therefore, a positive sig - CP cut point, SD standard deviation a ^ nal would still be detected if the sample contained up to Regression model: 5PL (auto) with weighting factor of 1/F 2 20,000 ng/mL anti-dostarlimab NAbs. Sensitivity was calculated using “average concentration + t 0.05, Z (one side) × SD” Selectivity Overall, 95% of the spiked samples (human cancer serum, The established LPC mean + t × SD = 339.3 ng∕mL + [2.33 × 83.4] 0.01,z (one side) hemolyzed normal serum, and lipemic normal serum) was 533.6 ng∕mL with a 1%false − negative rate. had mean ECL values greater than the plate-specific cut point, and 90% of the unspiked samples had mean ECL Assay precision values less than the plate-specific cut point (Table  4). Intra-assay precision was evaluated by running 6 inde- u Th s, the selectivity of the method met the acceptance pendent validation samples for each level (NC, LPC, and criteria with all 3 serum types: cancer serum, hemolyzed HPC) on a single plate. The %CV of the intra-assay preci - normal serum, and lipemic normal serum. sion was 7.7, 7.5, and 8.1 for the HPC, LPC (500 ng/mL), and NC respectively (Table 2). The %CV of the inter-assay Drug interference precision was 19.0, 23.9, 14.8, and 8.0 for the HPC, LPC-1 As shown in Table  5, all TSR-022–spiked NC samples (500 ng/mL), LPC-2 (533.6 ng/mL), and NC, respectively had ECL values below the cut point, and TSR-022–spiked (Table  3). Therefore, the intra-assay precision and inter- LPC samples had ECL values above the cut point. There - assay precision results met the specified criteria of %CV fore, as long as the drug removal step is performed, even ≤ 20% and %CV ≤ 25%, respectively (Gupta et al. 2011). high levels of TSR-022 (up to 1350 μg/mL) do not inter- fere with the signal for the LPC or NC. Hook effect Results in Table S2 show that the ECL signal of the NAb Assay robustness sample with the highest concentration (20,000 ng/mL) All robustness runs met the acceptance criteria was above the cut point. Even though the signal levels (Table  S3). Therefore, the assay was deemed robust with an incubation time ranging from 30 to 90 minutes per step: plate blocking, dostarlimab-NAb incubation, or sample incubation in MSD plate (Table S1). Table 2 Intra-assay precision Quality control # High‑positive Low‑positive Negative Stability control control control To assess the stability of NAbs, samples of HPC, LPC, 1 9.4 7.4 1.0 and NC were stored at 3 different conditions: 4 h at ambi - 2 9.2 7.3 0.9 ent temperature, 24 h at 2–8 °C, or subject to 6 freeze- 3 9.8 7.7 0.9 thaw cycles. All ECL signals of samples assessed for stability were within the established acceptance criteria 4 11.0 8.5 1.1 (Table  S4). Therefore, NAbs passed all the stability tests 5 9.2 7.7 1.0 performed. 6 10.4 8.7 1.1 Mean rECL 9.8 7.9 1.0 System suitability criteria %CV 7.7 7.5 8.1 The assay suitability criteria were used as part of the run %CV coefficient of variation, ECL electrochemiluminescence, rECL mean sample ECL value divided by mean plate negative control ECL value acceptance criteria to determine whether a run should Zhang et al. AAPS Open (2021) 7:8 Page 10 of 14 Table 4 Assay selectivity Sample type Lot NAb unspiked NAb spiked at LPC ECL %CV Plate CP Condition ECL %CV Plate CP Condition Cancer 1 1113 2.9 1573 < CP 2745 7.2 1573 > CP Cancer 2 1499 3.5 1573 < CP 2644 0.1 1573 > CP Cancer 3 1917 0.8 1573 > CP 2949 5.7 1573 > CP Cancer 4 1352 2.1 1573 < CP 2820 4.9 1573 > CP Cancer 5 1389 9.5 1573 < CP 2390 6.4 1573 > CP Cancer 6 1269 0.8 1573 < CP 3382 4.9 1573 > CP Cancer 7 1005 1.4 1573 < CP 2402 4.4 1573 > CP Cancer 8 998 7.4 1573 < CP 2362 1.2 1573 > CP Cancer 9 967 13.8 1573 < CP 1932 10.7 1573 > CP Cancer 10 836 11.9 1573 < CP 2478 1.3 1573 > CP Hemolyzed 1 1071 12.5 1573 < CP 2308 4.0 1573 > CP Hemolyzed 2 1144 0.8 1573 < CP 2745 19.9 1573 > CP Hemolyzed 3 893 9.0 1573 < CP 1717 3.4 1573 > CP Hemolyzed 4 802 0.4 1573 < CP 2066 10.7 1573 > CP Hemolyzed 5 553 2.0 1573 < CP 691 6.0 1573 < CP Lipemic 1 2561 6.3 1573 > CP 3486 5.7 1573 > CP Lipemic 2 1212 7.3 1573 < CP 2042 11.1 1573 > CP Lipemic 3 813 1.4 1573 < CP 1652 5.2 1573 > CP Lipemic 4 1525 6.0 1573 < CP 2758 1.0 1573 > CP Lipemic 5 1010 0.8 1573 < CP 1916 10.6 1573 > CP CP, cut point; %CV, coefficient of variation; ECL, electrochemiluminescence readout; LPC, low-positive control; NAb, neutralizing antibody Table 5 Drug interference TSR‑022 (μg/mL) NAb spiked at LPC level NAb unspiked ECL %CV Plate CP Condition ECL %CV Plate CP Condition 1350 14676 7.4 2484.5 > CP 2253 10.1 2484.5 < CP 450 11298 12.7 2484.5 > CP 2070 6.1 2484.5 < CP 150 11444 8.8 2484.5 > CP 1962 8.5 2484.5 < CP 0 13930 8.1 2484.5 > CP 2391 10.8 2484.5 < CP CP cut point, %CV coefficient of variation, ECL electrochemiluminescence readout, LPC low-positive control, NAb neutralizing antibody be rejected or accepted during validation stability testing Discussion and study sample analysis. The following system suitabil - Assay design ity criteria were obtained: Initially, a cell-based assay was chosen to determine if it would meet the acceptance criteria to function as the • NC range: rECL 0.76–1.24 NAb assay (Cong et  al. 2015). This cell-based biolumi - • LPC-1 range: rECL 1.82–11.01 nescent reporter assay was developed to quantify the • LPC-2 range: rECL 4.62–12.00 NAb to the therapeutic mAbs for either anti–PD-1 or • HPC range: rECL 5.11–18.36 anti–PD-L1 that block the PD-1/PD-L1 interaction. • NC < ACP < LPC < HPC PD-1 effector cells that express a nuclear factor of acti - vated T cells (NFAT)-luciferase reporter and a human PD-1 receptor were engineered from Jurkat T cells (Cong Analysis of all the validation runs showed that all QCs et  al. 2015). When the engineered Jurkat T cells were and NCs were within the acceptable range. Zhang  et al. AAPS Open (2021) 7:8 Page 11 of 14 cocultured with antigen-presenting cells that expressed detection. Furthermore, unlike most other cell-based T cell receptor (TCR) activator and PD-L1, activation of assays, which require only 1 type of cell line, the PD-1/ the NFAT pathway through TCR activator/TCR complex PD-L1 blockade bioassay required culturing, mainte- interaction would normally occur, leading to biolumines- nance, and preparation of 2 independent cell lines, fol- cence (Cong et  al. 2015). However, engagement of PD-1 lowed by co-culturing in an appropriate ratio to mimic and PD-L1 on the cells would block the NFAT pathway PD-1/PD-L1 function in vivo. Undoubtedly, this would and bioluminescence. Both anti–PD-1 and anti–PD-L1 potentially create more uncertainty leading to less sen- mAbs were able to block the PD-1/PD-L1 interaction, sitive and less reproducible results. resulting in the recovery of NFAT pathway signaling and The drug tolerance of this NAb assay was at least 250 bioluminescence (Cong et al. 2015). This assay, which can μg/mL with the implementation of the drug removal be used as a potency assay for anti–PD-1 drugs, was our process. This level of tolerance is based on the expo - starting point for NAb assay development. However, the sure level observed from the recommended therapeu- assay was labor intensive, had low tolerance for serum, tic dose (4 doses of 500 mg Q3W followed by 1000 mg and lacked sensitivity (data on file), which did not allow Q6W afterwards) established in the GARNET study. the cell-based reporter assay to move forward as a vali- The exposure evaluation was done at cycles 1, 4, 5, 8, dated anti-dostarlimab NAb assay. and 12, and only pre-dose samples were used for the The current trend for choosing a NAb assay format immunogenicity evaluation. During the treatment pro- includes using the therapeutic MOA as the driving selec- cess, the highest observed concentration, which was tion criterion and, in addition, relies on both assay perfor- confirmed ADA-positive through 3-tier ADA assays, mance characteristics and the risk of immunogenicity to was lower than 250 μg/mL. The drug tolerance of this the patient (Wu et al. 2016). When considering the MOA assay allows all samples to be evaluated with the low of the drug, both drug-target interaction and the struc- inconclusive rate. In the NAb assay validation, the cut tural characteristics need to be examined. Dostarlimab is point was determined using 60 individual human serum an antagonistic mAb that directly blocks the function of samples with the CPF of 1.18. The selection of the 60 the extracellular domains of PD-1, which are responsible cancer sera was balanced with respect to age, sex, and for the interaction with PD-L1 and signal transduction to cancer types. This CPF can be applied to different indi - the intra-cellular domains. Since dostarlimab blocks the cations without bias. The GARNET study includes mul - specific PD-1/PD-L1 interaction by binding to the PD-1 tiple patient populations, and in its NAb evaluation, it receptor, instead of binding at another site to exert its is unclear whether this pre-study cut point is suitable effect, a non–cell-based NAb assay is adequate for NAb for the population being studied (ClinicalTrials.gov detection (Wu et  al. 2016). Two other anti–PD-1 mAbs, 2016). Statistical determination of validation and in- nivolumab and pembrolizumab, also directly block the study immunogenicity cut points were performed by interaction of PD-1 with its ligands. Currently, non– the same statistician at Frontage Laboratory (Exton, PA) cell-based assays are used for the detection of NAbs to using the same procedures. An in-study cut point was nivolumab and pembrolizumab (van Vugt et  al. 2019; determined statistically. S/N ratio responses (also called Enrico et al. 2020). rECL) from 40 individual pre-dose baseline serum sam- ples from the GARNET study that were negative in the Assay performance and clinical relevance NAb assay were used for the cut point analysis. Out of A cell-based assay was evaluated before the white the 40 S/N ratios, no outliers were identified, and all paper on assay format assessment was published (Wu values were included in cut point determination. The et  al. 2016). Extensive optimization of the cell-based normality of the 40 S/N ratio responses was evaluated assay, including PD-L1 cell plating density, recovery using the Shapiro-Wilk test, and the test gave a P value of PD-1/PD-L1 co-culture or induction time, human = 0.0754, which is more than 0.01 (1%), indicating that serum concentration, cell culture plate type, luciferase the data set of 40 S/N ratio values was considered to be substrate volume and reaction time, and plate reader normally distributed. setting, were completed to improve the assay sensitiv- The in-study parametric CPF for the NAb was cal - ity. Under optimal assay conditions, the sensitivity in culated as 1.35 at a 1% false-positive error rate. In the the presence of 0.4 μg/mL of the drug was determined original NAb sample analysis based on the method vali- to be 12 μg/mL by screening 50 lots of human can- dation CPF of 1.18 at the 1% false-positive error rate for cer serum. Since this was far from the USA Food and the GARNET study, 64 of 135 confirmed ADA-positive Drug Administration recommended assay sensitiv- samples were determined to be NAb positive. Upon ity of 100 ng/mL, it was concluded that this assay was retrospective application of the higher in-study cut highly insensitive and was not serviceable for NAb point, 18 of the 64 samples switched from NAb positive Zhang et al. AAPS Open (2021) 7:8 Page 12 of 14 to NAb negative. Based on the 3-tier ADA assays, 13 tested and found to be acceptable. The performance of patients had treatment-emergent ADAs. Of these 13 the assay was independent of runs and analysts. With the patients, 7 patients were confirmed as NAb positive system’s suitability, stability, and drug tolerance, which and the other 6 were NAb negative using both the vali- covers the drug concentration range for the therapeu- dation cut point and the in-study cut point. Use of the tic dose established, this assay has been used to evalu- higher in-study NAb cut point did not change the num- ate NAbs to support the clinical trials that help guide the ber of patients with treatment-emergent ADAs who development of dostarlimab. were positive for NAbs, and thus would have no effect on the overall interpretation of the immunogenic- Abbreviations ity findings for this study using the tiered approach of ACP: Assay cut point; ADAs: Anti-drug antibodies; CPFs: Cut point factors; CV: 3-tier ADA assays and this NAb assay. Furthermore, Coefficient of variation; dMMR: Mismatch repair-deficient; ECL: Electrochemilu- minescence; ELISA: Enzyme-linked immunoassay; HPC: High-positive control; method sensitivity and drug tolerance should not be LPC: Low-positive control; mAb: Monoclonal antibody; MOA: Mechanism of greatly impacted by the lot of serum samples used for action; MSD: Meso Scale Diagnostics; NAbs: Neutralizing antibodies; NC: Nega- the establishment of cut point(s). The 60 lots of serum tive control; NFAT: Nuclear factor of activated T cells; PBS: Phosphate-buffered saline; PD-1: Programmed death 1; PD-L1: Programmed death ligand 1; PK: samples were selected to balance for cancer type, age, Pharmacokinetic; pNHS: Pooled normal human serum; SD: Standard deviation; and sex. In addition, only 40 pre-dose samples from S/N: Signal to noise; TCR : T cell receptor. the study were used to calculate the in-study cut point, which was not optimal. Method sensitivity and drug Supplementary Information tolerance should be evaluated by spiking in the pooled The online version contains supplementary material available at https:// doi. pre-dose samples of the study. This was not required org/ 10. 1186/ s41120- 021- 00039-w. here since the validation cut point can be further used due to limited impact as demonstrated in the above Additional file 1: Table S1. Validation of Assay’s Robustness. Table S2. Hook Eec ff t. Table S3. Assay’s Robustness. Table S4. Assay Stability analysis results. With the limited positive ADA data in the GARNET study, no clinically relevant impact was observed from PK, safety, or efficacy, which is consist - Acknowledgements Medical writing and editorial assistance, funded by GlaxoSmithKline ent with the observation that in currently approved ( Waltham, Massachusetts) and coordinated by Hasan Jamal, MSc, of Glaxo- anti–PD-1 mAbs, the presence of ADAs and NAbs SmithKline, were provided by Shannon Morgan-Pelosi, PhD, and Jennifer does not correlate with an impact on pharmacokinet- Robertson, PhD, of Ashfield MedComms, an Ashfield Health company (Mid- dletown, CT, USA). ics, pharmacodynamics, safety, or efficacy (Davda et al. 2019; Enrico et al. 2020). Authors’ contributions Allauthors provided substantial contributions to the conception or design of thework, or the acquisition, analysis, or interpretation of data for the work. Conclusions Allauthors drafted the work, revised it critically for important intellectualcon- tent, and provided final approval of the version to be published. Allauthors Because the clinical implications of ADAs and NAbs had access to full data and analyses presented in this manuscript. against dostarlimab remain unclear, both ADAs and NAbs to dostarlimab should be accurately detected Authors’ information Not applicable using an established assay, such as that described here (Enrico et  al. 2020). As has been demonstrated with Funding other anti–PD-1 mAbs, a validated non–cell-based com- Funding for this study was provided by GlaxoSmithKline (NCT02715284). Trademarks are owned by or licensed to the GSK group of companies. petitive ligand-binding assay is appropriate for detecting NAbs against dostarlimab with suitable precision, sensi- Availability of data and materials tivity, and selectivity (Enrico et al. 2020; Center for Drug GSK makes available anonymized individual participant data and associated documents from interventional clinical studies which evaluate medicines, Evaluation and Research 2017). Detection of ADAs, upon approval of proposals submitted to www. clini calst udyda tareq uest. com. especially NAbs, is extremely important for the evalua- To access data for other types of GSK sponsored research, for study docu- tion of efficacy and safety of biologic therapeutics. NAbs ments without patient-level data, and for clinical studies not listed, please submit an inquiry via the website. act by neutralizing biologic therapeutics; therefore, their presence may diminish drug efficacy. To this end, we Declarations have established an assay that can detect NAbs against dostarlimab, a humanized anti–PD-1 antibody. We have Competing interests determined the CPF for the detection of neutralizing Xiaolong Tom Zhang has nothing to disclose. Hong Chen has nothing to dis- close. Weiping Shao has nothing to disclose. Zhongping John Lin has nothing anti-dostarlimab antibodies in human serum at a 1% to disclose. Murad Melhem is an employee of GlaxoSmithKline. Sharon Lu is an false-positive rate. The assay’s precision, sensitivity, hook employee of Scholar Rock and was affiliated with GlaxoSmithKline at the time effect, selectivity, robustness, and drug interference were of the study analyses. Zhang  et al. AAPS Open (2021) 7:8 Page 13 of 14 Author details review of registration trials and future considerations. J Immunother WuXi AppTec, 288 Fute Zhong Road, Waigaoqiao Free Trade Zone, Shang- Cancer 6(1):8 hai 200131, China. Alliance Pharma, 17 Lee Boulevard, Malvern, PA 19355, Gunn GR, Sealey DCF, Jamali F, Meibohm B, Ghosh S, Shankar G (2016) From USA. Frontage Laboratories, 700 Pennsylvania Drive, Exton, PA 19341, USA. the bench to clinical practice: understanding the challenges and GlaxoSmithKline, 1000 Winter Street, Waltham, MA 02451, USA. uncertainties in immunogenicity testing for biopharmaceuticals. Clin Exp Immunol 184(2):137–146 Received: 4 June 2021 Accepted: 13 October 2021 Gupta S, Devanarayan V, Finco D, Gunn GR III, Kirshner S, Richards S et al (2011) Recommendations for the validation of cell-based assays used for the detection of neutralizing antibody immune responses elicited against biological therapeutics. J Pharm Biomed Anal 55(5):878–888 Jolicoeur P, Tacey RL (2012) Development and validation of cell-based assays for the detection of neutralizing antibodies to drug products: a practical References approach. Bioanalysis. 4(24):2959–2970 Ahn BC, Pyo KH, Xin CF, Jung D, Shim HS, Lee CY et al (2019) Comprehensive Kaufman HL, Russell J, Hamid O, Bhatia S, Terheyden P, D’Angelo SP et al (2016) analysis of the characteristics and treatment outcomes of patients with Avelumab in patients with chemotherapy-refractory metastatic Merkel non-small cell lung cancer treated with anti-PD-1 therapy in real-world cell carcinoma: a multicentre, single-group, open-label, phase 2 trial. practice. J Cancer Res Clin Oncol 145(6):1613–1623 Lancet Oncol 17(10):1374–1385 Andre T, Berton D, Curigliano G, Ellard S, Perez JMT, Arkenau HT et al (2021) Khozin S, Miksad RA, Adami J, Boyd M, Brown NR, Gossai A et al (2019) Real- Safety and efficacy of anti-PD-1 antibody dostarlimab in patients with world progression, treatment, and survival outcomes during rapid adop- mismatch repair-deficient (dMMR) solid cancer: results from the GARNET tion of immunotherapy for advanced non-small cell lung cancer. Cancer. study. J Clin Oncol 39(suppl 3):abstr 9 125(22):4019–4032 Ansell SM, Lesokhin AM, Borrello I, Halwani A, Scott EC, Gutierrez M et al Krishna M, Nadler S (2016) Immunogenicity to biotherapeutics - the role of (2015) PD-1 blockade with nivolumab in relapsed or refractory Hodgkin’s anti-drug immune complexes. Front Immunol 7:21 lymphoma. N Engl J Med 37(4):311–319 Levene H (1960) Robust tests for the equality of variances. In: Oaklin I (ed) Antonia SJ, Villegas A, Daniel D, Vicente D, Murakami S, Hui R, PACIFIC Inves- Contributions to probability and statistics. Stanford University Press, Palo tigators et al (2017) Durvalumab after chemoradiotherapy in stage III Alto, pp 278–292 non-small-cell lung cancer. N Engl J Med 377(20):1919–1929 Liao K, Sikkema D, Wang C, Chen K, DeWall S, Lee TN (2012) Development and Apolo AB, Ellerton JA, Infante JR, Agrawal M, Gordon MS, Aljumaily R et al validation of a cell-based SEAP reporter assay for the detection of neutral- (2017) Updated efficacy and safety of avelumab in metastatic urothelial izing antibodies against an anti-IL-13 therapeutic antibody. J Immunol carcinoma (mUC): pooled analysis from 2 cohorts of the phase 1b Javelin Methods 375(1–2):258–263 solid tumor study. J Clin Oncol 35(15 suppl):4528 Mire-Sluis AR, Barrett YC, Devanarayan V, Koren E, Liu H, Maia M et al (2004) Bloem K, Hernández-Breijo B, Martínez-Feito A, Rispens T (2017) Immuno- Recommendations for the design and optimization of immunoassays genicity of therapeutic antibodies: monitoring antidrug antibodies in a used in the detection of host antibodies against biotechnology products. clinical context. Ther Drug Monit 39(4):327–332 J Immunol Methods 289:1–16 Center for Drug Evaluation and Research (2017) Approval package for: Oaknin A, Ellard SL, Leath C III, Moreno V, Kristeleit R, Guo W et al (2018) application number 125514Orig1s014. https:// www. acces sdata. fda. Preliminary safety, efficacy, and PK/PD characterization from GARNET, a gov/ drugs atfda_ docs/ nda/ 2019/ 12551 4Orig 1s014. pdf. Accessed 4 phase 1 clinical trial of the anti–PD-1 monoclonal antibody, TSR-042, in Feb 2021. patients (pts) with recurrent or advanced MSI-H endometrial cancer. Ann ClinicalTrials.gov (2016) Study of TSR-042, an anti-programmed cell death-1 Oncol 29(suppl8):viii334 receptor (PD-1) monoclonal antibody, in participants with advanced Pineda C, Hernandez GC, Jacobs IA, Alvarez DF, Carini C (2016) Assessing the solid tumors (GARNET ): NCT02715284. https:// clini caltr ials. gov/ ct2/ show/ immunogenicity of biopharmaceuticals. BioDrugs. 30:195–206 NCT02 715284. Accessed 6 May 2020. https:// clini caltr ials. gov/ ct2/ show/ Robert C, Ribas A, Wolchok JD, Hodi FS, Hamid O, Kefford R et al (2014) NCT02 715284? term= 02715 284& draw= 2& rank=1 Anti-programmed-death-receptor-1 treatment with pembrolizumab in Cong M, Cheng Z, Karassina N, Grailer J, Hartnett J, Cosby N, Fan F et al (2015) ipilimumab-refractory advanced melanoma: a randomised dose-compar- Advertorial: novel bioassay to assess PD-1/PD-L1 therapeutic antibodies ison cohort of a phase 1 trial. Lancet. 384(9948):1109–1117 in development for immunotherapy. Genet Eng News 35(10) https:// Seiwert TY, Burtness B, Mehra R, Weiss J, Berger R, Eder JP et al (2016) Safety www. genen gnews. com/ magaz ine/ 247/ adver torial- novel- bioas say- to- and clinical activity of pembrolizumab for treatment of recurrent or assess- pd-1- pd- l1- thera peutic- antib odies- in- devel opment- for- immun metastatic squamous cell carcinoma of the head and neck (KEY- other apy/. Accessed 14 Aug 2020 NOTE-012): an open-label, multicentre, phase 1b trial. Lancet Oncol Davda J, Declerck P, Hu-Lieskovan S, Hickling TP, Jacobs IA, Chou J et al (2019) 17(7):956–965 Immunogenicity of immunomodulatory, antibody-based, oncology Shankar G, Devanarayan V, Amaravadi L, Barrett YC, Bowsher R, Finco-Kent D therapeutics. J Immunother Cancer 7:105 et al (2008) Recommendations for the validation of immunoassays used El-Khoueiry AB, Sangro B, Yau T, Crocenzi TS, Kudo M, Hsu C et al (2017) for detection of host antibodies against biotechnology products. J Pharm Nivolumab in patients with advanced hepatocellular carcinoma (Check- Biomed Anal 48(5):1267–1281 Mate 040): an open-label, non-comparative, phase 1/2 dose escalation Shapiro SS, Wilk MB (1965) Analysis of variance test for normality (complete and expansion trial. Lancet. 389(10088):2492–2502 samples). Biometrika. 52:591–611 Enrico D, Paci A, Chaput N, Karamouza E, Besse B (2020) Antidrug antibodies US Food and Drug Administration. (2021a) FDA grants accelerated approval against immune checkpoint blockers: impairment of drug efficacy or to dostarlimab-gxly for dMMR endometrial cancer. https:// www. fda. indication of immune activation? Clin Cancer Res 26(4):787–792 gov/ drugs/ drug- appro vals- and- datab ases/ fda- grants- accel erated- European Medicines Agency (2021) Jemperli. https:// www. ema. europa. eu/ en/ appro val- dosta rlimab- gxly- dmmr- endom etrial- cancer. Accessed 24 medic ines/ human/ EPAR/ jempe rli. Accessed 24 May 2020 May 2021. Fehrenbacher L, Spira A, Ballinger M, Kowanetz M, Vansteenkiste J, Mazieres US Food and Drug Administration. (2021b) FDA grants accelerated approval to J, POPLAR Study Group et al (2016) Atezolizumab versus docetaxel for dostarlimab-gxly for dMMR advanced solid tumors. https:// www. fda. gov/ patients with previously treated non-small-cell lung cancer (POPLAR): drugs/ resou rces- infor mation- appro ved- drugs/ fda- grants- accel erated- a multicentre, open-label, phase 2 randomised controlled trial. Lancet. appro val- dosta rlimab- gxly- dmmr- advan ced- solid- tumors. Accessed 16 387(10030):1837–1846 Sept 2021. Garcês S, Demengeot J (2018) The immunogenicity of biologic therapies. Curr van Vugt MJH, Stone JA, De Greef RHJMM, Snyder ES, Lipka L, Turner DC et al Probl Dermatol 53:37–48 (2019) Immunogenicity of pembrolizumab in patients with advanced Gong J, Chehrazi-Raffle A, Reddi S, Salgia R (2018) Development of PD-1 and tumors. J Immunother Cancer 7(1):212 PD-L1 inhibitors as a form of cancer immunotherapy: a comprehensive Zhang et al. AAPS Open (2021) 7:8 Page 14 of 14 Weber JS, D’Angelo SP, Minor D, Hodi FS, Gutzmer R, Neyns B et al (2015) Wu Y, Akhgar A, Li J, Yu B, Chen C, Lee N et al (2018) Selection of a ligand- Nivolumab versus chemotherapy in patients with advanced mela- binding neutralizing antibody assay for benralizumab: comparison with noma who progressed after anti-CTLA-4 treatment (CheckMate 037): an antibody-dependent cell-mediated cytotoxicity (ADCC) cell-based a randomised, controlled, open-label, phase 3 trial. Lancet Oncol assay. AAPS. 20(3):49 16(4):375–384 Wu B, Chung S, Jiang XR, McNally J, Pedras-Vasconcelos J, Pillutla R et al (2016) Publisher’s Note Strategies to determine assay format for the assessment of neutralizing Springer Nature remains neutral with regard to jurisdictional claims in pub- antibody responses to biotherapeutics. AAPS J 18(6):1335–1350 lished maps and institutional affiliations.

Journal

AAPS Open – Springer Journals

Published: Nov 11, 2021

Keywords: Dostarlimab; Endometrial cancer; Neutralizing antibodies; Immunotherapy; Competitive ligand-binding assay

There are no references for this article.