AAPS Journal最新文献

Ocrelizumab (OCREVUS®) is a humanized anti-CD20 monoclonal antibody approved for the treatment of adult patients with relapsing multiple sclerosis (RMS) and primary progressive multiple sclerosis (PPMS). Here, we discuss the strategic and technical considerations needed to develop a robust antibody-dependent cellular cytotoxicity (ADCC)-based neutralizing antibody (NAb) assay to detect anti-ocrelizumab NAb in patients enrolled in the ocrelizumab registered clinical trials. The NAb detection assay consisted of a two-tier assay that included a screening assay and a confirmation assay. In the screening assay, patient samples were analyzed in the presence of ocrelizumab. Samples that tested positive in the screening assay were subsequently analyzed in the confirmatory assay where another anti-CD20 mAb, obinutuzumab, was replaced by ocrelizumab, to verify NAb specificity. Both assays utilized MEC-2 cells, a chronic B cell leukemia cell line, pre-labeled with calcein AM as the target cells, and natural killer (NK) cells engineered to stably express Fc gamma receptor IIIa_ F158 as effector cells. Both cell lines were prepared to be thaw-and-use cells. The NAb assay measures fluorescence from the calcein AM released into the assay media upon the lysis of target cells by ADCC in the presence of ocrelizumab or obinutuzumab. Our validated NAb assay showed a relative sensitivity of 743 ng/mL and can detect 1500 ng/mL of a surrogate positive control antibody in the presence of 1500 ng/mL ocrelizumab. This ADCC assay is the first reported NAb assay that directly measures target cell lysis by using thaw-and-use target and effector cells simultaneously.

Cell and gene therapies have demonstrated impressive therapeutic efficacy in various human diseases. Nevertheless, cellular immune response directed against these therapeutic agents is an obstacle for achieving long-lasting clinical efficacy. Therefore, it is crucial to develop robust assays to accurately monitor cellular immunogenicity towards these therapies. Enzyme-linked immunospot (ELISpot) assay is one of the primarily used methods for measuring cellular immune response in clinical programs, which requires isolation of the peripheral blood mononuclear cells (PBMCs). The quality of this clinical material is one of the most critical factors that impact the robust assessment of cellular immune responses. The optimal blood sample processing conditions, however, remain poorly understood. In this study, we examined the impact of blood sample processing time on the performance characteristics of ELISpot to measure antigen-specific cellular responses. Blood samples that were processed after overnight delay resulted in a loss of ELISpot signals. We subsequently optimized several parameters of sample processing, and successfully recovered ELISpot signals for the blood samples that are processed within 32 h. Furthermore, several mitigation strategies were employed that would potentially address the impact of granulocyte contamination on detection of antigen-specific cellular responses. Our investigation provides an extension of sample processing window for clinical studies and is significant for resolving the logistical challenge of whole blood sample shipment for timely PBMC preparation in cell/gene therapy clinical studies.

PF-07257876 is a bispecific antibody being developed for the treatment of certain advanced or metastatic solid tumors. To support clinical development of PF-07257876, neutralizing antibody (NAb) assays were developed as part of a tiered immunogenicity testing approach. Because PF-07257876 targets both CD47 and PD-L1, determination of domain specificity of a NAb response may provide additional insight relating to PK, efficacy, and safety. Due to limitations of functional cell systems, two cell-based binding assays were developed using electrochemiluminescence to detect domain-specific NAb. While both NAb assays utilized a cell-based binding approach and shared certain requirements, such as sensitivity and tolerance to potentially interfering substances, the development of each assay faced unique challenges. Among the hurdles encountered, achieving drug tolerance while preserving domain specificity for CD47 proved particularly challenging. Consequently, a sample pretreatment procedure to isolate NAb from potentially interfering substances was necessary. The sample pretreatment procedure developed was based on a bead-extraction and acid dissociation (BEAD) approach. However, the use of the standard BEAD approach with whole drug to capture NAb resulted in loss of NAb detection under certain circumstances. Specifically, mock samples containing a mixture of NAb positive controls against both binding domains of the bispecific antibody produced false-negative results in the cell-based binding assay. An adaptation made to the standard BEAD approach restored domain-specific NAb detection, while also contributing to an assay sensitivity of 1 µg/mL in the presence of a clinically relevant drug tolerance level of up to 400 µg/mL.

Poly lactic-co-glycolic acid (PLGA) microparticles have been formulated to allow the sustained release of numerous drugs, including antibodies. It is well-known that antibodies are susceptible to chemical and physical stress; therefore, it is necessary to be loaded on PLGA microparticles under mild conditions. In the present study, we constructed cationic porous PLGA microparticles that could be electrostatically adsorbed with infliximab as a model antibody. Cationic porous PLGA microparticles were prepared using the double emulsion method by adding polyethyleneimine and ammonium bicarbonate. After antibody loading, surface pores closure was achieved by mild heating. The size of the optimized formulation was approximately 5 μm, exhibiting a positive charge. The loaded antibody was gradually released from the formulation over 56 days. Based on a tumor necrosis factor (TNF)-α inhibition assay, the released infliximab maintained its pharmacological activity. Collectively, we successfully loaded antibodies into PLGA microparticles while maintaining activity and demonstrating long-acting properties.

Process analytical technology (PAT) in late-stage drug product development is typically used for real-time process monitoring, in-process control, and real-time release testing. In early research and development (R&D), PAT usage is limited as the manufacturing scale is relatively small with frequent changes and only a few batches are produced on an annual basis. However, process understanding is critical at early R&D in order to identify process and formulation boundaries, so PAT applications could be particularly useful in early-stage R&D. For oral solid dosage form, conventional HPLC-based content uniformity (CU) methods with sampling of 3 tablets per stratified sampling location in early R&D are typically not sufficient to identify these manufacturing process boundaries and temporal profile. Here, we report a screening CU method based on a multivariate model using transmission Raman spectroscopy (TRS) data on a phase-appropriate calibration set of only 16 tablets. This initial model was used for multiple pre-GMP development batches to provide critical information about blend uniformity and content uniformity (CU). In this work, the precision of the TRS method was evaluated; multiple spectral preprocessing approaches were compared regarding their effects on measurement precision as well as their ability to mitigate the photo bleaching effects during precision experiments. Overall, the TRS-based CU method was much faster than a traditional HPLC-based method allowing a much larger number of tablets to be screened. This larger number of analyzed tablets enabled the processes boundaries and temporal changes in CU to be identified while providing proper statistical assurance on product quality.

Qualifying critical reagents in ligand binding assays by parallel testing of current and candidate reagent lots is recommended by regulatory agencies and industry groups, but specific guidance on the format of reagent qualification experiments is limited. Equivalence testing is a statistically sound approach that is consistent with the objective of critical reagent qualification. We present power analysis for equivalence regions ranging from 1.25- to 1.5-fold multiples of the GM ratio (centered on 1) of current and candidate lots, over a range of assay variability from 5 to 30% coefficient of variation (CV). A 1.25-fold equivalence region can be tested using 6 to 12 plates per lot for assays with up to 15% CV but is not practical for more variable assays. For these assays, wider equivalence regions are justified so long as care is taken to avoid assay drift and the assay remains suitable for the intended use. The equivalence test method is illustrated using historical data from passing and failing reagent qualification experiments. Simulation analysis was performed to support the design of qualification experiments using 6, 12, or 18 plates per lot over a broad range of assay variability. A challenge in implementing the equivalence test approach is selecting an appropriate equivalence region. Equivalence regions providing 90% power using 12 plates/lot were consistent with 1.5σ bounds, which are recommended for equivalence testing of critical quality attributes of biosimilars.

Multidrug resistance (MDR1) and breast cancer resistance protein (BCRP) play important roles in drug absorption and distribution. Computational prediction of substrates for both transporters can help reduce time in drug discovery. This study aimed to predict the efflux activity of MDR1 and BCRP using multiple machine learning approaches with molecular descriptors and graph convolutional networks (GCNs). In vitro efflux activity was determined using MDR1- and BCRP-expressing cells. Predictive performance was assessed using an in-house dataset with a chronological split and an external dataset. CatBoost and support vector regression showed the best predictive performance for MDR1 and BCRP efflux activities, respectively, of the 25 descriptor-based machine learning methods based on the coefficient of determination (R²). The single-task GCN showed a slightly lower performance than descriptor-based prediction in the in-house dataset. In both approaches, the percentage of compounds predicted within twofold of the observed values in the external dataset was lower than that in the in-house dataset. Multi-task GCN did not show any improvements, whereas multimodal GCN increased the predictive performance of BCRP efflux activity compared with single-task GCN. Furthermore, the ensemble approach of descriptor-based machine learning and GCN achieved the highest predictive performance with R² values of 0.706 and 0.587 in MDR1 and BCRP, respectively, in time-split test sets. This result suggests that two different approaches to represent molecular structures complement each other in terms of molecular characteristics. Our study demonstrated that predictive models using advanced machine learning approaches are beneficial for identifying potential substrate liability of both MDR1 and BCRP.

The identification and removal of host cell proteins (HCPs) from biologic products is a critical step in drug development. Despite recent improvements to purification processes, biologics such as monoclonal antibodies, enzyme replacement therapies, and vaccines that are manufactured in a range of cell lines and purified using diverse processes may contain HCP impurities, making it necessary for developers to identify and quantify impurities during process development for each drug product. HCPs that contain sequences that are less conserved with human homologs may be more immunogenic than those that are more conserved. We have developed a computational tool, ISPRI-HCP, that estimates the immunogenic potential of HCP sequences by evaluating and quantifying T cell epitope density and relative conservation with similar T cell epitopes in the human proteome. Here we describe several case studies that support the use of this method for classifying candidate HCP impurities according to their immunogenicity risk.

Machine learning (ML) approaches have been applied to predicting drug pharmacokinetic properties. Previously, we predicted rat unbound brain-to-plasma ratio (Kpuu,brain) by ML models. In this study, we aimed to predict human Kpuu,brain through animal ML models. First, we re-evaluated ML models for rat Kpuu,brain prediction by using trendy open-source packages. We then developed ML models for monkey Kpuu,brain prediction. Leave-one-out cross validation was utilized to rationally build models using a relatively small dataset. After establishing the monkey and rat ML models, human Kpuu,brain prediction was achieved by implementing the animal models considering appropriate scaling methods. Mechanistic NeuroPK models for the identical monkey and human dataset were treated as the criteria for comparison. Results showed that rat Kpuu,brain predictivity was successfully replicated. The optimal ML model for monkey Kpuu,brain prediction was superior to the NeuroPK model, where accuracy within 2-fold error was 78% (R² = 0.76). For human Kpuu,brain prediction, rat model using relative expression factor (REF), scaled transporter efflux ratios (ERs), and monkey model using in vitro ERs can provide comparable predictivity to the NeuroPK model, where accuracy within 2-fold error was 71% and 64% (R² = 0.30 and 0.52), respectively. We demonstrated that ML models can deliver promising Kpuu,brain prediction with several advantages: (1) predict reasonable animal Kpuu,brain; (2) prospectively predict human Kpuu,brain from animal models; and (3) can skip expensive monkey studies for human prediction by using the rat model. As a result, ML models can be a powerful tool for drug Kpuu,brain prediction in the discovery stage.

Historically, the biopharmaceutical industry has used titer to characterize the magnitude of an anti-drug antibody (ADA) response. While reporting levels of antibodies in terms of titer is generally understood and accepted by regulatory and medical communities, titer values are inherently variable given the multiple serial dilutions and reporting a value either directly before or interpolated at the assay cut point on the lower plateau of the assay curve range. Using S/N is an appealing alternative approach to titer as it simplifies analysis with less dilutions, significantly reducing testing, time, and resources and provides a more precise value potentially differentiating low-level ADA responses. Current bridging electrochemiluminescence (ECL) ADA assays using Meso Scale Discovery (MSD) platform are also significantly more sensitive and drug tolerant with wider assay ranges compared to historic ELISA platforms; therefore, ADA response based on S/N may help differentiate and identify those ADA samples that are more likely to be clinically relevant. Bococizumab is a humanized monoclonal antibody targeting proprotein convertase subtilisin-kexin type 9 (PCSK9), which reduces plasma levels of low-density lipoprotein (LDL) cholesterol. Bococizumab was discontinued during Phase 3 clinical development based in part on the high rate of ADA and wide variation in LDL cholesterol responses among patients. The impact of anti-bococizumab antibodies on pharmacokinetic (PK) and pharmacodynamic (PD) endpoints was originally assessed using titer. Retrospective analysis of anti-bococizumab ADA responses using S/N ratios illustrates that S/N is an acceptable alternative to titer for characterizing the magnitude of ADA response and interpretation of clinically relevant ADA.