Lina Loo†, Shannon Harris†Mark Milton†, Meena†, Wibke Lembke†, Flora Berisha†, Sylvie Bertholet†, Francis Dessy†, Robert Dodge†, Xiaodong Fang†, Michele Fiscella†, Fabio Garofolo†, Boris Gorovits†, Soumi Gupta†, Vibha Jawa†, Akiko Ishii-Watabe†, Brian Long†, Yanmei Lu†, Timothy Mack†, Kristina McGuire†, Katrina Nola
Abstract
The 15th edition of the Workshop on Recent Issues in Bioanalysis (15th WRIB) was held on 27 September to 1 October 2021. Even with a last-minute move from in-person to virtual, an overwhelmingly high number of nearly 900 professionals representing pharma and biotech companies, contract research organizations (CROs), and multiple regulatory agencies still eagerly convened to actively discuss the most current topics of interest in bioanalysis. The 15th WRIB included 3 Main Workshops and 7 Specialized Workshops that together spanned 1 week in order to allow exhaustive and thorough coverage of all major issues in bioanalysis, biomarkers, immunogenicity, gene therapy, cell therapy and vaccines. Moreover, in-depth workshops on biomarker assay development and validation (BAV) (focused on clarifying the confusion created by the increased use of the term “Context of Use – COU”); mass spectrometry of proteins (therapeutic, biomarker and transgene); state-of-the-art cytometry innovation and validation; and, critical reagent and positive control generation were the special features of the 15th edition. This 2021 White Paper encompasses recommendations emerging from the extensive discussions held during the workshop, and is aimed to provide the bioanalytical community with key information and practical solutions on topics and issues addressed, in an effort to enable advances in scientific excellence, improved quality and better regulatory compliance. Due to its length, the 2021 edition of this comprehensive White Paper has been divided into three parts for editorial reasons. This publication (Part 3) covers the recommendations on TAb/NAb, Viral Vector CDx, Shedding Assays; CRISPR/Cas9 & CAR-T Immunogenicity; PCR & Vaccine Assay Performance; ADA Assay Comparability & Cut Point Appropriateness. Part 1A (Endogenous Compounds, Small Molecules, Complex Methods, Regulated Mass Spec of Large Molecules, Small Molecule, PoC), Part 1B (Regulatory Agencies’ Inputs on Bioanalysis, Biomarkers, Immunogenicity, Gene & Cell Therapy and Vaccine) and Part 2 (ISR for Biomarkers, Liquid Biopsies, Spectral Cytometry, Inhalation/Oral & Multispecific Biotherapeutics, Accuracy/LLOQ for Flow Cytometry) are published in volume 14 of Bioanalysis, issues 9 and 10 (2022), respectively.
Keywords:
| Abbreviations | |
| AAV: | Adeno-associated virus |
| Ab: | Antibody |
| ACE: | Affinity capture elution |
| ADA: | Anti-drug antibody |
| ADCC: | Antibody-dependent cellular cytotoxicity |
| Anti-id: | Anti-idiotypic |
| ARC: | ADA reagent complexes |
| BAb: | Binding antibody |
| BAV: | Biomarker assay validation |
| BCR: | B cell receptor |
| BLA: | Biologics license application |
| BMV: | Bioanalytical method validation |
| BsAb: | Bispecific antibody |
| BTM: | Blood transcription modules |
| CAR-T: | Chimeric antigen receptor T cell |
| CDC: | Complement-dependent cytotoxicity |
| CDx: | Companion diagnostic |
| cGMP: | Current good manufacturing practices |
| CHO: | Chinese hamster ovary |
| CIC: | Circulating immune complexes |
| CK: | Cellular kinetics |
| CLIA: | Clinical Laboratory Improvement Amendments |
| Companion diagnostics: | A companion diagnostic device can be an in vitro diagnostic device, testing kit or imaging tool that provides information that is essential for the safe and effective use of a corresponding therapeutic product. |
| COU: | Context of use |
| CP: | Cut point |
| CP-ARC: | Cut point-ADA reagent complex |
| CRISPR: | Clustered regularly interspaced short palindromic repeats |
| CRISPR-Cas9: | It stands for clustered regularly interspaced short palindromic repeats and CRISPR-associated protein 9. By delivering the Cas9 nuclease complexed with a synthetic guide RNA (gRNA) into a cell, the cell’s genome can be cut at a desired location, allowing existing genes to be edited, removed and/or new ones added. |
| CRO: | Contract research organization |
| CTD: | Common technical document |
| CV: | Coefficient of variation |
| DCA: | Domain competition assays |
| DDA: | Domain detection assays |
| dPCR: | Digital polymerase chain reaction |
| ddPCR: | Droplet digital polymerase chain reaction |
| DNA: | Deoxyribonucleic acid |
| DoE: | Design of experiments |
| DT: | Drug tolerance |
| ECLA: | Electrochemiluminescence assay |
| ECD: | Extra cellular domain |
| ELISA: | Enzyme-linked immunosorbent assay |
| ELISpot: | Enzyme-linked immune absorbent spot |
| FFP: | Fit for purpose |
| FIH: | First in human |
| FIX: | Factor IX |
| FPR: | False positive rates |
| FSC/SSC: | Forward scatter/Side scatter |
| GCP: | Good Clinical Practices |
| gDNA: | Genomic DNA |
| GTx: | Gene therapy |
| HAMA: | Human anti-mouse antibodies |
| hFIX: | Human Factor IX |
| HLA: | Human leukocyte antigen |
| HMW: | High molecular weight |
| hFVIII: | Human Factor VIII |
| IC: | Immune complex |
| ICS: | Intracellular cytokine staining |
| IDE: | Investigational device exemption |
| IFN-γ: | Type II interferon |
| IMPD: | Investigational Medicinal Product Dossier |
| IND: | Investigational new drug |
| ISI: | Integrated Summary of Immunogenicity |
| ISR: | Incurred sample reanalysis |
| IVD: | In vitro device |
| KOL: | Key opinion leader |
| LBA: | Ligand binding assay |
| LCM: | Life cycle management |
| LCMS: | Liquid chromatography mass spectrometry |
| LDT: | Laboratory developed test |
| LIMS: | Laboratory Information Management System |
| LLOQ: | Lower limit of quantitation |
| LOD: | Limit of detection |
| MAA: | Marketing authorization application |
| mAb: | Monoclonal antibody |
| MHC: | Major histocompatibility complex |
| MIQE: | It stands for the minimum information for publication of quantitative real-time PCR experiments. MIQE guidelines describe the minimum information necessary for evaluating qPCR experiments and target the reliability of results to help ensure the integrity of the scientific literature, promote consistency between laboratories, and increase experimental transparency. |
| MOA: | Mechanism of action |
| mRNA: | Messenger RNA |
| MSR: | Minimum significant ratio |
| NAb: | Neutralizing antibody |
| NHP: | Non-human primate |
| NTC: | No template control |
| pAb: | Polyclonal antibody |
| PAI: | Pre-approval inspection |
| PBMC: | Peripheral blood mononuclear cell |
| PC: | Positive control |
| PCR: | Polymerase chain reaction |
| PD: | Pharmacodynamic |
| PHA: | Phytohemagglutinine |
| PK: | Pharmacokinetics |
| PMC: | Postmarketing commitment |
| PPV: | Positive predictive value |
| PRNT: | Plaque reduction neutralization test |
| QC: | Quality control |
| qPCR: | Quantitative polymerase chain reaction |
| RCV: | Replication competent virus |
| RG: | Reference genes |
| RIN: | RNA integrity number |
| RNA: | Ribonucleic acid |
| RNP: | Ribonucleoprotein |
| ROA: | Route of administration |
| RT: | Reverse transcription |
| rVLP: | Recombinant virus-like particles |
| SAP: | Statistical analysis plan |
| scFv: | Single-chain variable fragment |
| SEC: | Size exclusion chromatography |
| sgRNA: | Single guide RNA |
| SNR: | Signal to noise ratio |
| TAb: | Total antibody |
| TCR: | T cell receptor |
| TE: | Target engagement |
| Tfh: | T follicular helper |
| TI: | Transduction inhibition |
| Transduction inhibition: | The inhibition of the transduction of cells by serum or other matrices in a cell-based assay. The inhibition may be caused by antibodies, low molecular weight drugs, or proteins present in the sample. |
| TLF: | Tables, listing and figures |
| Vector shedding: | The dissemination of viral vector released outside the treated subject via excreta (e.g., urine and feces), and secreta (e.g., saliva, semen, sweat). |
| VRNT: | Virus reduction neutralization test |
| WRIB: | Workshop on Recent Issues in Bioanalysis |
Index Part 3
SECTION 1 – Gene Therapy, Cell Therapy and Vaccines
Hot Topics & Consolidated Questions Collected from the Global Bioanalytical Community
Discussions, Consensus and Conclusions
SECTION 2 – Immunogenicity of Biotherapeutics
Hot Topics & Consolidated Questions Collected from the Global Bioanalytical Community
Discussions, Consensus and Conclusions
Introduction
The 15th edition of the Workshop on Recent Issues in Bioanalysis (15th WRIB) was held on 27 September to 1 October 2021. Even with a last minute move from in-person to virtual, an overwhelmingly high number of nearly 900 professionals representing pharma and biotech companies, contract research organizations (CROs), and multiple regulatory agencies still eagerly convened to actively discuss the most current topics of interest in bioanalysis. The 15th WRIB included three Main Workshops and seven Specialized Workshops that together spanned 1 week in order to allow exhaustive and thorough coverage of all major issues in bioanalysis, biomarkers, immunogenicity, gene therapy, cell therapy and vaccines.
Moreover, in-depth workshops on biomarker assay development and validation (BAV) (focused on clarifying the confusion created by the increased use of the term “Context of Use – COU”); mass spectrometry of proteins (therapeutic, biomarker and transgene); state-of-the-art cytometry innovation and validation; and, critical reagent and positive control generation were the special features of the 15th edition.
As in previous years, this year’s WRIB continued to gather a wide diversity of international, industry opinion leaders and regulatory authority experts working on both small and large molecules as well as gene and cell therapies to facilitate sharing and discussions focused on improving quality, increasing regulatory compliance, and achieving scientific excellence on bioanalytical issues.
The active contributing chairs included: Dr. Eugene Ciccimaro (BMS), Dr. Anna Edmison (Health Canada), Dr. Fabio Garofolo (BRI), Dr. Swati Gupta (AbbVie), Dr. Shannon Harris (HilleVax), Dr. Carrie Hendricks (Sanofi), Ms. Sarah Hersey (BMS), Dr. Steve Keller (AbbVie), Dr. Lina Loo (Vertex), Dr. Mark Ma (Alexion), Dr. Joel Mathews (Ionis), Dr. Meena (Stoke), Dr. Manoj Rajadhyaksha (Alexion), Dr. Ragu Ramanathan (Vertex), Dr. Susan Spitz (Incyte), Dr. Dian Su (Mersana), Dr. Matthew Szapacs (Abbvie), Dr. Albert Torri (Regeneron), Dr. Jian Wang (Crinetics), Dr. Jan Welink (EU EMA), and Dr. Yuling Wu (AstraZeneca).
The participation of major and influential regulatory agencies continued to grow at the 15th WRIB during its traditional Interactive Regulators’ sessions including presentations and panel discussions on:
- Regulated Bioanalysis and BMV Guidance/Guidelines: Dr. Seongeun Julia Cho (US FDA), Dr. Arindam Dasgupta (US FDA), Dr. Anna Edmison (Health Canada), Dr. Elham Kossary (WHO), Mr. Gustavo Mendes Lima Santos (Brazil ANVISA), Dr. Sam Haidar (US FDA), Dr. Sandra Prior (UK MHRA), Dr. Mohsen Rajabi Abhari (US FDA), Dr. Diaa Shakleya (US FDA), Dr. Catherine Soo (Health Canada), Dr. Nilufer Tampal (US FDA), Mr. Stephen Vinter (UK MHRA), Dr. Yow-Ming Wang (US FDA), Drs Jan Welink (EU EMA), and Dr. Jinhui Zhang (US FDA)
- Biotherapeutic Immunogenicity, Gene Therapy, Cell Therapy and Vaccines: Dr. Nirjal Bhattarai (US FDA), Dr. Elana Cherry (Health Canada), Dr. Isabelle Cludts (UK MHRA), Dr. Heba Degheidy (US FDA), Dr. Soma Ghosh (US FDA), Dr. Akiko Ishii-Watabe (Japan MHLW), Dr. Susan Kirshner (US FDA), Dr. Kimberly Maxfield (US FDA), Dr. Joao Pedras-Vasconcelos (US FDA), Dr. Mohsen Rajabi Abhari (US FDA), Dr. Vijaya Simhadri (US FDA), Dr. Dean Smith (Health Canada), Dr. Therese Solstad (EU EMA/Norway NoMA), Dr. Daniela Verthelyi (US FDA), Dr. Meenu Wadhwa (UK MHRA), Ms. Leslie Wagner (US FDA), Dr. Günter Waxenecker (Austria AGES), Dr. Haoheng Yan (US FDA), and Dr. Lucia Zhang (Health Canada)
- Biomarkers/CDx and BAV Guidance/Guidelines: Mr. Abbas Bandukwala (US FDA), Dr. Soma Ghosh (US FDA), Dr. Shirley Hopper (UK MHRA), Dr. Kevin Maher (US FDA), Dr. Yoshiro Saito (Japan MHLW), and Dr. Yow-Ming Wang (US FDA)
All the traditional “working dinners” attended by both industry key opinion leaders (KOL) and regulatory representatives were held in a virtual format this year, and the extensive and fruitful discussions from these “working dinners” together with the lectures and open panel discussions amongst the presenters, regulators and attendees culminated in consensus and recommendations on items presented in this White Paper.
A total of 66 recent issues (‘hot’ topics) were addressed and distilled into a series of relevant recommendations. Presented in the current White Paper is the background on each issue, exchanges, discussions, consensus and resulting recommendations.
Due to its length, the 2021 edition of this comprehensive White Paper has been divided into three parts for editorial reasons. This publication covers Part 3 recommendations.
Part 1 – Volume 14 Issue 9 (May)
Regulated Mass Spectrometry of Large Molecules (six topics)
| 1. | Regulated Bioanalysis/ICH M10 of Large Molecules by Mass Spectrometry | ||||
| 2. | Internal Standard Selection in Hybrid Assays in Regulated Bioanalysis/ICH M10 | ||||
| 3. | Regulatory Feedback on Nanomedicine Bioanalysis by Mass Spectrometry | ||||
| 4. | Bioanalytical Mass Spectrometry Strategies for CRISPR Quantification | ||||
| 5. | Extracellular Vesicle Bioanalysis by Mass Spectrometry | ||||
| 6. | Immunocapture Platform Considerations for Intact Mass LCMS | ||||
Endogenous Compounds & Complex Methods (five topics)
| 1. | Chiral Methods for Method Development and BMV | ||||
| 2. | ICH M10 Section 7.1 for Endogenous Compound Quantification | ||||
| 3. | Tissue Analysis, Rare Matrices and Atypical Sample Collection in Regulated Bioanalysis | ||||
| 4. | Quantitation of Intracellular Disposition of Oligonucleotides and Sensitivity/Specificity Challenges | ||||
| 5. | Recent Developments of Urinary Endogenous Compounds and Fit-for-Purpose Validation | ||||
Regulated Bioanalysis for Small Molecule & Point of Care (six topics)
| 1. | Dealing with GLP, GCP and GCLP Frameworks in Regulated Bioanalysis | ||||
| 2. | Importance of Incurred Sample Stability in Regulated Bioanalysis | ||||
| 3. | Challenges when Changing Platforms (LBA to LCMS) in Regulated Bioanalysis | ||||
| 4. | Patient-Centric Approaches and Point of Care in Regulated Bioanalysis | ||||
| 5. | Regulatory Standards to Perform Bioanalysis in China | ||||
| 6. | Bioanalytical Challenges for Oncology Drug Development | ||||
Mass Spectrometry of Proteins (six topics)
| 1. | Hybrid Assays to Quantify Therapeutics Proteins | ||||
| 2. | PTM/Glycosylation Analysis for Biomarkers and Biotherapeutics | ||||
| 3. | Hybrid Assays to Quantify Protein Biomarkers | ||||
| 4. | Hybrid Assays for Target Engagement Assessment | ||||
| 5. | Quantification of ADA by Hybrid Assays | ||||
| 6. | Hybrid Assays to Quantify Transgene Proteins | ||||
Input from Regulatory Agencies on Regulated Bioanalysis & BMV
Input from Regulatory Agencies on Immunogenicity, Biomarkers, Gene Therapy, Cell Therapy and Vaccines
Part 2 – Volume 14 Issue 10 (May)
Biomarkers & CDx Development & Validation (nine topics)
| 1. | Liquid Biopsy: Challenges and Opportunities with Extracellular Vesicles | ||||
| 2. | ISR for Biomarker Assays, Parallelism & Biomarker Assay Validation Guidance | ||||
| 3. | Clinical Biomarkers as Surrogate Endpoints or for Patient Segmentation | ||||
| 4. | Quality Oversight of CLIA Laboratories for Companion Diagnostics | ||||
| 5. | Emerging Trends and Impact on Diagnostic Development | ||||
| 6. | Breath & Airway Biomarker Determination | ||||
| 7. | High Sensitivity Platforms for Biomarkers and Companion Diagnostics | ||||
| 8. | Exploratory and Target Engagement Biomarker Assays | ||||
| 9. | PBMC Sample Collection for Pharmacodynamic Biomarkers | ||||
Cytometry Validation & Innovation (eight topics)
| 1. | Recent Developments in Flow Cytometry Validation in a Bioanalytical Lab | ||||
| 2. | Evaluation of Accuracy for Flow Cytometry in Regulated Laboratories | ||||
| 3. | Sensitivity Determination in Flow Cytometry Validation | ||||
| 4. | Clinical Biomarker Development, Validation and Interpretation by Cytometry | ||||
| 5. | Spectral Cytometry Ultra-High Order/Dimensional Assays in Clinical Applications | ||||
| 6. | Imaging Cytometry Quantitative Analysis of Target Engagement | ||||
| 7. | Mass Cytometry in Clinical Biomarkers and “Clinical Trial Compatibility” | ||||
| 8. | Multivariate Analytical Techniques and Multiparameter Flow Cytometry | ||||
LBA Regulated Bioanalysis, Critical Reagents & Positive Controls (nine topics)
| 1. | Bioanalytical Challenges for Inhalation and Oral Delivery of Biologics | ||||
| 2. | Free, Bound, Total, Active, Monoactive, Biactive, and Multiactive PK Assays | ||||
| 3. | Implementing Free/Total PK Assays in Regulated Bioanalysis | ||||
| 4. | Multi-Domain Biotherapeutic PK Assays in Regulated Bioanalysis | ||||
| 5. | Bioanalytical Challenges to Study the Biodistribution of Biotherapeutics | ||||
| 6. | Advanced Approaches in Critical Reagent Selection for PK Assays | ||||
| 7. | Challenges with Positive Control Generation for ADA Assays | ||||
| 8. | Critical Reagent Assay Comparability | ||||
| 9. | Novel Critical Reagent Modalities: “Thinking out of the Box” | ||||
Part 3 – Volume 14 Issue 11 (June)
Gene Therapy, Cell Therapy & Vaccines (nine topics)
| 1. | TAb/NAb & Anti-Viral Vector Antibody Companion Diagnostic Assays | ||||
| 2. | Viral Vector Shedding Assays | ||||
| 3. | Viral Vector Gene Therapy Immunogenicity & Pre-Existing Immunity | ||||
| 4. | CRISPR/Cas9 Immunogenicity & Bioanalytical Challenges | ||||
| 5. | CAR-T Immunogenicity & Cellular Kinetics | ||||
| 6. | qPCR/ddPCR Assay Performance | ||||
| 7. | Analyte Stability in Vaccine Serology Assays | ||||
| 8. | Vaccine Critical Reagent Management and Bridging | ||||
| 9. | Vaccine Bioanalytical Assays & Immune Monitoring | ||||
Immunogenicity of Biotherapeutics (eight topics)
| 1. | NAb Assays – Drug and Target Interference | ||||
| 2. | ADA Cut Points – Appropriateness and False Positive Rates | ||||
| 3. | Circulating Immune Complexes – ADA/Drug Complexes | ||||
| 4. | ADA Assay Comparability | ||||
| 5. | Integrated Summary of Immunogenicity Harmonization | ||||
| 6. | China NMPA Immunogenicity Guidance | ||||
| 7. | Multi-Domain Biotherapeutics & Bispecific Antibody Immunogenicity | ||||
| 8. | Biosimilar ADA Assay Validation & Harmonization | ||||
SECTION 1 – Gene Therapy, Cell Therapy & Vaccines
Lina Loo1, Shannon Harris2, Mark Milton18, Meena3, Wibke Lembke15, Flora Berisha4, Sylvie Bertholet5, Francis Dessy6, Robert Dodge7, Xiaodong Fang8, Michele Fiscella9, Fabio Garofolo10, Boris Gorovits11, Soumi Gupta12, Vibha Jawa13, Akiko Ishii-Watabe14, Brian Long12, Yanmei Lu16, Timothy Mack13, Kristina McGuire17, Katrina Nolan19, Luying Pan20, Bernd Potthoff21, Shobha Purushothama22, Dean Smith24, Therese Solstad25, Ivo Sonderegger26, Frank Taddeo20, Shabnam Tangri27, Leslie Wagner23, Bonnie Wu4 & Yuanxin Xu28
Authors are presented in alphabetical order of their last name, with the exception of the first 5 authors who were session chairs, working dinner facilitators and/or major contributors.
The affiliations can be found at the beginning of the article.
HOT TOPICS & CONSOLIDATED QUESTIONS COLLECTED FROM THE GLOBAL BIOANALYTICAL COMMUNITY
The topics detailed below were considered as the most relevant “hot topics” based on feedback collected from the 14th WRIB attendees. They were reviewed and consolidated by globally recognized opinion leaders before being submitted for discussion during the 15th WRIB. The background on each issue, discussions, consensus and conclusions are in the next section and a summary of the key recommendations is provided in the final section of this manuscript.
TAb/NAb & Anti-Viral Vector Antibody Companion Diagnostic Assays
Can a standardized anti-viral vector serotype binding antibody (BAb) and neutralizing Ab (NAb) be used to evaluate assay sensitivity, or should an anti-viral serotype positive control (PC) be used as a reference standard? Are both required or is one sufficient? How does humoral response as measured by total antibody (TAb) or NAb impact adeno-associated virus (AAV) transduction? Which assay (TAb, NAb) is best suited for the assessment of pre-existing reactivity to AAV? For the more sensitive TAb assay, would a confirmatory/specificity step help? Which assay should be considered as a potential companion diagnostic (CDx)? When is a CDx essential? What if a CDx is not available? When might the FDA forgo the requirement for a CDx? Can a CDx be used for more than one product? Is a CDx needed for a clinical trial?
Viral Vector Shedding Assays
How can shedding assays be put into practice? Could prior data be used? Is monitoring necessary until three consecutive results are below the lower limit of quantitation (LLOQ)? When considering matrices for the shedding assays, should blood be used? Can empty capsids be tracked? Why consider shedding evaluations for non-replicating viruses? Should lifestyle restrictions be set up for patients treated with AAV-based therapy based on serotype, route of administration (ROA), or dose? Which matrices should be used in relation to the ROA and dose? How many subjects should be evaluated and what time points should be selected, including termination of testing? Is plasmid DNA or vector the more appropriate reference material for standard calibrators? How should recovery be assessed during sample analysis? What analytical platforms and reporting units should be selected, including for matrices that do not have quantifiable endogenous DNA?
Viral Vector Gene Therapy Immunogenicity & Pre-Existing Immunity
What are the important determinants of pre-existing immunity that impact transduction? Are low ex vivo neutralization titers clinically relevant? What is the impact on overcoming pre-existing immunity of empty or light capsids in the vector preparation? What are the best animal models to study the role of pre-existing immunity and what interventions should be used to overcome it? Is there a way to induce a pre-existing type of immune response in animals such as rabbits? Is effective redosing with an AAV-based gene therapy possible? What are the most promising approaches so far? What are the multiple components of AAV gene therapies that may elicit immunogenicity? What are the pros and cons of assay methodologies used to detect pre-existing immunity to AAV? What are the immunogenicity considerations for clinical monitoring? What is the potential impact of AAV-specific cellular immune response on gene therapy safety and efficacy? If the transgene protein is expressed as an intracellular product, is it necessary to assess humoral response to the transgene protein? Are total binding anti-drug antibody (ADA) assay results adequate for patient pre-screening? What are the factors contributing to persistent transgene expression? Which assays are essential and which are not to support gene therapy programs?
CRISPR/Cas9 Immunogenicity & Bioanalytical Challenges
What are some of the unique challenges while assessing the immunogenicity to Cas9? What are the critical reagent characterization needs for evaluating the immunogenicity of Cas9? What are the current regulatory expectations for gene editing therapeutics regarding exposure/biodistribution data, measurement of both total and active ribonucleoprotein (RNP) complex and testing of tissue biopsies in clinical studies as well as the concern with off-target effects? As a consequence of T cell immune response to chimeric antigen receptor T cells (CAR-T), has the FDA seen data on safety? What are the options for the assessment of safety using animal models given that not all animal models are relevant? What is the most appropriate bioanalytical strategy for quantifying cellular therapies?
CAR-T Immunogenicity & Cellular Kinetics
What is the optimal platform to assess ADA and NAb response? When should one evaluate cytotoxic T cell response? What do regulators consider acceptable CAR components for a screening LBA: source antibody, single-chain variable fragment (scFv), extracellular part (scFv + linker + hinge), full construct including intracellular parts? What are suitable strategies to address cut point (CP) calculations for high incidence of pre-existing immunogenicity, e.g., use of immunoglobulin depleted sera, selected low responder sera, patient specific cut points?
qPCR/ddPCR Assay Performance
What are some of the pros and cons of using RNA versus DNA standards for the quantitation of messenger RNA (mRNA) copy numbers of a transgene encoding a transcription regulator using reverse transcription-quantitative polymerase chain reaction (RT-qPCR) to support investigational new drug (IND)-enabling studies? What are the approaches for producing RNA reference material and how is this critical reagent characterized? How should standards and quality control (QC) samples be prepared? What are the parameters to be evaluated and how should acceptance criteria be set? What are the relevant assay validation parameters and considerations when using qPCR/RT-qPCR assays for vaccines? What are the sources of pathogen reference standard material to support limit of detection (LOD)/LLOQ assessments for vaccine PCR and RT-qPCR assays? What are the considerations in bridging/calibration of pathogen reference standards for vaccine PCR and RT-qPCR assays? Are updates needed to recent recommendations on qPCR assay validation for gene therapy, viral shedding and biodistribution [28]? What is the current status of regulatory requirements for qPCR validation (e.g., when should the requirement of LLOQ = 50 vector copy numbers/μg DNA be applied)? What are the best practices for assay validation, including acceptance criteria and reporting units, considering the stage of drug development? What assay performance controls and QCs are relevant? What are the assay life cycle management considerations? What are the considerations for the validation of qPCR versus droplet digital PCR (ddPCR)? Should there be probe redundancy for qPCR-based tests? What is the cost and benefit to increased PCR targets to minimize the likelihood of false negatives? Are quantitative or qualitative tests needed to distinguish high versus low viral loads or antibodies since most tests seem to be binary?
Analyte Stability in Vaccine Serology Assays
What is required to demonstrate short- and long-term analyte stability for antibodies in serum? Could the field leverage data from what has been previously demonstrated? What is the level of assay characterization needed (i.e., should assays be fit for purpose [FFP], qualified, or validated)? What are the best practices for trending assay performance and managing out of trend events?
Vaccine Critical Reagent Management & Bridging
How are the complementary tools used to ensure assay stability (e.g., bridging of critical reagents, inclusion of control samples, control trending analyses, proficiency panel) integrated into a coherent system for managing long-term assay performance? Can acceptance criteria be harmonized? How can realistic alarm thresholds be set for control trending tools used to detect assay trends before the out-of-specification rate increases? Can consensus be reached to replace traditional schemes (e.g., “assays are set-up in phase I, qualified for phase II and validated for phase III” or “assay used to support exploratory endpoints are set up, those to support secondary endpoints are qualified and primary endpoints are validated”) with the following language: assay performance characteristics critical to the intended use are appropriately assessed and controlled? Why is taking a reagent life-cycle management (LCM) approach critical? What are some bridging approaches between different lots of reagents in vaccine assays? Who are the key stakeholders in setting these acceptance criteria?
Vaccine Bioanalytical Assays & Immune Monitoring
When using new, modern assay platforms, should we bridge to gold standard platforms? When is it appropriate to engage with the regulators on acceptance of new platforms? What is the extent of bridging necessary when data is not compared across clinical phases? How can assay development and qualification be streamlined? Do assays need to be qualified for Phase I and validated for Phase IIb if data are included in the submission? What are some approaches for including exploratory/translational research in mainstream Phase I clinical trials? What is the appropriate level of assay development for clinical exploratory/translational research readouts? How can integrated computational analyses be harnessed when there is limited powering of data due to usual small Phase I studies?
DISCUSSIONS, CONSENSUS & CONCLUSIONS
TAb/NAb & Anti-Viral Vector Antibody Companion Diagnostic Assays
Development of gene therapies using viral vectors requires a suite of assays to evaluate potential immune responses, including responses to the viral vector as well as to the transgene protein [29,30]. These assays may include an assay to measure transduction inhibition (TI), i.e., the ability of a biological matrix to inhibit the transduction of the viral vector to target cells in vitro. Such assays were previously called neutralizing antibody or NAb assays; however, recently this terminology was changed in recognition of the fact that there may be components other than antibodies that can modulate viral transduction [31].
Pre-existing antibodies to the viral vectors may have a negative impact on the safety and efficacy of viral vector-based gene therapies. Hence detection of these antibodies is important. The prevalence of pre-existing antibodies varies with capsid serotype, geographic region and age of the population [32,33], although there may or may not be a meaningful clinical impact depending on the dosing of the vector template serotype, antibody titer, and ROA. Regardless of pre-existing antibody status, AAV vector administration results in seroconversion to significant and durable AAV-specific antibody responses. Although the clinical impact of antibodies that develop following dose administration may or may not be relevant for the initial dose, the durability of the high titer is considered sufficient to prevent successful repeat dose administrations with the same or similar vector capsids [34].
Pre-existing antibodies represent prior viral exposure and may have a negative impact on safety and efficacy of the administered gene therapy. The antibody response may be specific towards one or more AAV serotypes due to overall cross-reactivity. Sponsors may choose to exclude patients with pre-existing antibodies above a certain titer, either TAb or NAb, to the viral vector out of caution. The use of pre-treatment immune suppressive strategies to overcome the need for pre-screening was discussed. There was consensus amongst the expert panel that prophylactic regimens with immune suppressive/modulatory treatments would not mitigate the impact of pre-existing antibodies. However, they may diminish treatment boosted or emergent immunogenicity and have broader impact on immune functions. Additionally, strategies that physically or enzymatically remove antibodies for use of empty viral capsids to neutralize pre-existing antibodies prior to dosing were also discussed and not recommended due to limited data supporting their use. It is not well understood why some of these clinical mitigation regimens work for some patients but not for others. The ROA is also an important consideration in evaluating the impact of pre-existing antibodies as a patient inclusion/exclusion criterion [28]. For example, intrathecal or ocular administration-based clinical trials have not excluded patients based on pre-existing antibodies [35,36]. Systemic administration of gene therapies will require a scientifically driven, fit for purpose strategy.
There still needs to be clarity in terms of whether the assays currently used to evaluate pre-existing antibodies to the different AAV serotypes meet the criteria of CDx based on the FDA guidance. CDx assays (also known as in vitro devices [IVD] or testing kits) provide information on patient management and also on the safe and effective use of an associated therapeutic or drug. The CDx assay will be used once the therapeutic has been approved for marketing. Recent FDA guidance strongly recommends considering development of a CDx when pre-existing antibodies to AAV are to be used as an exclusion criterion. Sponsors are encouraged to share their pre-enrollment and testing strategy with the FDA during the early stages of development [37]. Similar regulations in other countries also exist. Currently, the expectation is that the CDx and the therapeutic in development should both be approved contemporaneously, as needed. The need for a CDx versus a Clinical Laboratory Improvement Amendments (CLIA)-based assay used mainly for clinical diagnosis in the context of gene therapy for clinical trial eligibility needs to be carefully considered and debated for this specific use. The development of such assays would require availability of relevant robust and specific reagents to support patient care, and patients who may receive clinically meaningful benefit are not unfairly excluded.
In the event a CDx is not available at the time of marketing authorization, sponsors should engage with regulators early on during their product development to develop alternative tests such as laboratory developed tests (LDTs), which may be used in the absence of an approved CDx.
At the onset of a study, there is a risk determination that should be submitted to the regulatory agencies. Based on the risk (significant or non-significant) the agency will determine if an investigational device exemption (IDE) needs to be submitted for their review for those studies that are of significant risk. Given that patient safety is of paramount importance, the regulatory agencies may place a clinical trial on hold if there are doubts about the analytical validity of an assay being used to enroll patients in the trial. The expert panel discussed the expectations for an IVD-grade assay for screening for pre-existing antibodies and enabling patient inclusion/exclusions enrollment. Several logistical and technical challenges were identified to implement such titer (TAb or NAb) cut-off to include or exclude assays at earlier stages of development during clinical trials, including lack of sufficient relevant sample volume or technical challenges with implementing an IVD or CDx assay with a complex format. When a sponsor believes that a regulated clinically validated assay poses a non-significant risk (i.e., an incorrect test result does not pose a potential for serious risk to subjects in a trial), submitting a justification for this position will aid in the review of IND materials considering a totality of evidence approach.
There is a need for standardized assays that can support the development of multiple serotypes for different sponsors. While such assays can be used to support a clinical trial, they may not be implemented as a CDx. The regulators, however, are interested in these platform approaches, and this continues to be a topic for discussion. It was also concluded that a CDx for one AAV gene therapy (GTx) may be used as a LDT for the development of another AAV GTx of the same serotype. Given the cross-reactive nature of antibodies among serotypes, especially between viral capsid of a specific serotype and drug product, an antibody method for a generic capsid for a specific serotype may be sufficient instead of for the drug product; this would allow the use of the same assay to test pre-existing antibodies (in-treatment naive samples) and post-treatment samples for immunogenicity.
Another area of active discussion with the regulatory agencies is the monitoring of inter-laboratory performance of the CDx if multiple labs are expected to perform the test. It is anticipated that this will be a topic of discussion in the near future.
Some of the key inputs that still require confirmation following prior industry/regulators’ discussions [21,25,28] are whether the pre-existing antibody response impacts the AAV transduction and whether a TAb or NAb assay is better suited or equivalent to measure these responses. The following case studies illustrate key points from the discussion.
In the first case study, an AAV8 was administered to mice with pre-existing NAb and BAb; the ADA-complexed capsid was taken up by both liver parenchymal and non-parenchymal cells although there was no transgene expression. In studying how BAbs influence systemic AAV-mediated gene delivery parameters, including transgene expression levels, vector persistence in the systemic circulation, and vector immunogenicity in pre-immunized animals, it was noted that BAbs have a markedly different effect on AAV vector mediated liver transduction and tissue biodistribution compared to NAbs. Mice with BAbs showed an improved transduction and a unique vector biodistribution profile [38], indicating that the NAb was a better predictor of AAV vector transduction than BAb. The other case involved Factor IX (FIX) therapy using an AAV5 vector conducted in non-human primates (NHP), where pre-existing anti-AAV5 NAb did not seem to impact transduction based on the measured human FIX (hFIX) protein levels. Therefore, a clinical study was conducted in patients (n = 3) with pre-existing anti-AAV5 NAb. The NAb were measured using a more sensitive luciferase reporter in a cell-based assay. However, no relationship was observed between the presence of pre-existing anti-AAV5 NAb and therapeutic efficacy [39], indicating that low titer NAb may not necessarily have clinical relevance. Hence there may not be a need to constantly refine the NAb assay to improve sensitivity. Similarly, an AAV5 serotype vector encoding human Factor VIII (hFVIII) was administered to NHP with pre-existing immunity measure as both TAb and TI. Discordant animals in which only low titer TI was detected in the absence of a TAb signal (TI+/TAb-) achieved hFVIII plasma protein concentrations similar to control animals with no pre-existing immunity. Only the n = 5 animals where TAb was detected as pre-existing (TI+/TAb+) showed markedly diminished efficacy, suggesting subjects with low levels of TI in the absence of detectable TAb may achieve a meaningful pharmacodynamic effect [34]. Finally, a study correlating TAbs with NAbs using AAV9 vector did not initially demonstrate correlation until the introduction of a confirmatory step in the TAb assay, after which there was good correlation between the two assays [40]. Sensitivity of the NAb assay is dependent on the format, choice of reporter (green fluorescent protein versus luciferase), and minimum required dilution (MRD). The impact of NAb on transduction is dependent on the AAV serotype, tissue of interest, target and ROA.
Despite recent advances, there is limited experience related to the assays used for the detection of pre-existing antibody responses. The criterion for patient inclusion/exclusion could be based on sensitivity of detection (generally with a TAb assay format) or clinical relevance (typically with a Nab assay). NAb assays, in general, are surrogate reporter gene cell-based assays with higher variability and matrix interference than TAb methods. Therefore, TAb assays are most often implemented for ease of use, sensitivity, high throughput and clinical relevance, as observed by association of TAbs with immunotoxicities. The need for a NAb assay specific for the viral vector and transgene protein is evaluated on a case by case basis. It was recommended to consider the pros and cons of both TAb and NAb assay formats as shown in Table 1 during the development of a bioanalytical strategy.
| LBA-Based TAb Assays | Cell-Based NAb Assays | |
|---|---|---|
| Pros | • Detects all Abs that bind to AAV capsid antigen; can be serotype specific | |
| • Detects all antibodies that could impact transduction by different mechanisms • Help interpret association with immunotoxicities • High throughput, automation friendly • More robust and sensitive • Easy to implement in a pre-screening setting, CDx friendly | • Assumed to be better predictors of transduction outcomes • Clinically relevant due to inhibition of transduction although a surrogate reporter gene based cell line is typically used | |
| Cons | • High sensitivity can lead to high prevalence of ADA positive results • Not all ADA antibodies detected are neutralizing | • Prolonged assay development (selection or engineering of cell lines with relevant receptor and suitable reporter genes) • Labor intensive, low throughput, high variability, matrix interference-inhibition resulting in high % inhibition CP, with potential false positive results due to non-NAb inhibitors • Off target effects of BAb mediated through FcR or complement receptor mediated uptake into reticuloendothelial cells is not captured |
A recent publication by Gorovits et al. [31], illustrates some of the considerations when developing a TAb assay. Ideally, the format should be such that it can detect all isotypes (primarily IgG and IgM) against the capsid. The formats typically used are direct assays, indirect assays, or bridging assays using enzyme-linked immunosorbent assays (ELISA) or electrochemiluminescence assays (ECLA). The assays may be performed in a homogeneous or step-wise manner and use anti-isotype antibodies as the detection reagent. A confirmatory assay using depletion with the vector, pre-treatment with capsid resin, or removal of IgGs, may also be performed. It should be noted that some sponsors will not perform the screening and confirmatory assays but instead analyse all samples using a titration assay. This approach is relevant when there is a high incidence of “positive” samples (e.g., all post-dose samples will likely confirm positive) and time is of the essence to screen the patients for inclusion in the clinical trial for treatment.
Careful consideration should be given to the critical reagents. Commercially available PCs may be used, which can be monoclonal antibodies with specific serotype specificity, a non-human control screened for cross reactivity or cross-reactive polyclonal antibody tested against the relevant disease matrix. The assay negative controls could be pooled seronegative sera or serum samples immunodepleted for specific and non-specific cross-reactivity. Capture reagents could be the therapeutic; however, empty capsids, recombinant virus-like particles (rVLPs) or capsid-derived specific domains or peptides could be used if the therapeutic is not available in sufficient quantity for assay development and testing. Given that these assays are in place early in the development process, consideration should be given to critical reagent bridging in pace with therapeutic development. The use of generic and commercially available AAV capsid as the antigen (critical reagent) for the TAb assay instead of using the drug product was also discussed. This would be relevant as anti-AAV antibodies have a good degree of cross-reactivity among different natural AAV serotypes and drug products and TAb responses (pre-existing and post-dose immunogenicity) in patients are polyclonal, use of a generic AAV method (instead of drug viral template specific method) would be most efficient.
Viral Vector Shedding Assays
Recombinant AAV is an established therapeutic modality for delivery of gene therapies but can pose potential safety and environmental concerns. Vector shedding, defined as the dissemination of viral vector released into the environment from the treated subject via excreta (e.g., urine and feces) or secreta (e.g., saliva, semen, sweat), is perceived as one of these potential concerns based on the theoretical possibility for transmission to untreated individuals although some of the AAV capsid containing transgene is generally not replication competent. Regulatory guidance documents [41–43] set very clear expectations with regards to the assessments of shedding, requiring that studies be conducted to measure potential transmission using robust shedding data to inform the shedding profile and the environmental risk assessment. However, sponsors are encouraged to keep in mind that these shedding studies are not mass balance studies. The need for infectivity assays can also impact the study design particularly if more frequent or long-term follow up of the replication competent virus (RCV) is needed [44]. Infectivity assays are typically cell-based and subject to matrix interference, with poor precision and sensitivity. PCR-based methods are becoming more common.
The assessment of vector shedding as an environmental risk is also a current requirement for replication incompetent and non-pathogenic vectors such as AAV [25,41]. Even with ROAs where the probability of virus shedding is low (e.g., subretinal administration), viral shedding should be assessed with additional focus on in vivo systemically administered therapies [45]. However, the recent guidance does not list viral shedding follow-up studies, indicating that regulators may not always require this assessment. With certain viruses (e.g., oncolytic viruses and herpes simplex virus) that have a higher risk of shedding live infectious virus, cell-based infectivity assays may be required to understand and mitigate potential exposure to non-treated individuals [25]. The exception is for ex vivo administered lentiviral vectors, for which no infectivity assay is required [41]. Overall, virus shedding is an important area to study and understand when developing a GTx [41,46] and the need for shedding assays when using replication incompetent and non-pathogenic viruses such as AAV [42] and the practicality of implementing the requirements of the FDA guidance [41] were key discussion topics. However, it should be noted that there are no examples of shed replication-incompetent viral vector-based gene therapies having ever resulted in inadvertent transduction in another person or animal.
In terms of the need to perform shedding studies for replication-competent and non-pathogenic viruses, at this time there is insufficient real-world data to obtain consensus on obviating the need for these assays, nor is there evidence that shedding is an issue. Sponsors are encouraged to keep in mind that the current regulatory requirements are conservative due to an abundance of caution and limited clinical experience, reflect the current thinking and are not intended to be static. Sponsors are encouraged to engage with the regulators and utilize data collected over the course of product development as justification for not conducting future shedding studies.
Given that virus shedding is transient, at levels much lower than the dosed AAV and potentially influenced by pre-existing AAV antibodies [47], the need to conduct virus shedding assessments pose unique bioanalytical challenges in terms of considering the analytical platforms, matrices and timing of sample collection that are dependent on the type of viral vector, vector tropism and ROA [41]. The current industry practice is to use qPCR/ddPCR-based methods for shedding assays and/or cell-based infectivity assays, if assay performance supports its use [48–50]; the challenges associated with the latter have been previously described [21]. There is currently no regulatory guidance on the selection of analytical methodologies and how to conduct method characterization, validation and clinical sample testing for shedding assays; industry best practices provide guidance on the considerations and parameters that are important to characterize and validate qPCR methods [21,25,28,48].
There was consensus that the most commonly used matrices for shedding assessment are urine, nasopharyngeal swabs, saliva, tears (for local eye delivery), and semen (adult male subjects within age group only). It is well understood and accepted that the collection of samples to assess shedding may be inconvenient to patients and assay sensitivity in these matrices vary. Experience from some GTx developers indicates that once the therapy is in the cellular compartment, there is little evidence in the plasma compartment; hence blood is not a typical matrix of choice for vector shedding but is a critical matrix for biodistribution to help interpret vector shedding data. There can be long-term shedding or detection of genomes in the peripheral blood, but this is typically cell-associated (red blood cells for months, peripheral blood mononuclear cell (PBMC) for years) and not found free in the plasma [51]. When developing qPCR methods, the type of consent needed to collect samples, sample collection aspects due to degradation by endogenous enzymes, and analytical platform sensitivity should be considered. With regards to sample collection (excretory or secretory), there was consensus that missing time points are a reality for these studies and the key challenge is in making these shedding studies as patient-friendly as possible while collecting the maximum amount of data. Regulators are willing to accept these types of deviations to the study plan based on limited impact on the available data.
The need to track empty capsids without transgene or light capsids that may contain unintended and poorly defined nucleic acid sequences derived from the manufacturing process was discussed. It was concluded that current manufacturing processes generally keep this impurity assessment to ≤5% even if in some cases empty AAV capsids can be up to ∼80% of total viral particles from batch to batch. Inter-patient variability may make it hard to standardize the number of empty capsids, and the bioanalytical methods currently used to detect AAV capsids do not have the specificity to distinguish empty capsids from AAV capsids and fully transgene-loaded AAV capsids, nor will they ever be able to do so for gene therapies that use endogenous AAVs. In addition, if the intended use of tracking the empty capsid is a strategy to decrease pre-existing antibodies as an adsorption method, there is no evidence that this results in a better outcome for patients in comparison to plasmapheresis and/or intravenous immune globulin (human immunoglobulin) therapy and is not recommended. Finally, there is no reason that there should be clearance differences between the empty capsid and the one containing the transgene as it is unlikely that the transgene contents change the ability of the capsid to be taken up by cells.
Viral Vector Gene Therapy Immunogenicity & Pre-Existing Immunity
Treatment with GTx introduces many foreign proteins that have the potential to elicit immune responses. Hence, like other biologics, it is important to perform and update an immunogenicity risk assessment. Although neither FDA nor EMA have introduced guidance on immunogenicity assessments that are specific for viral vector GTx, it is widely accepted that current immunogenicity guidance documents [52,53] are broadly applicable [31]. Other aspects of immunogenicity assessments agreed to as industry best practices include banking of samples, conduct of the immunogenicity assays at a single laboratory and evaluating the impact of immunogenicity on efficacy and adverse events.
Often, GTx are preferably evaluated in seronegative patients first, the scientific understanding being that anti-AAV antibodies resulting from prior exposure to AAV may limit efficacy and impact safety of AAV-based gene therapies. Pre-existing antibodies to AAV could impact transduction by neutralizing the vector (preventing binding of the AAV to its receptor on target cells) or by altering biodistribution (formation of large immune complexes and subsequent clearance of the AAV via the reticuloendothelial system). Additionally, treating patients with pre-existing antibodies may lead to immune complex formation, complement activation, and other potentially severe immunotoxicities. The assessment of pre-existing antibodies and its use to determine inclusion/exclusion criteria may be most relevant for systemic routes of administration.
Prevalence of pre-existing antibodies to AAV varies by serotype, age and geography, with the AAV5 serotype showing the lowest rate of prevalence compared to other serotypes such as AAV6, AAV2, AAV8, etc. [54]. The titers of the pre-existing antibodies are magnitudes lower than those associated with post-treatment immune response [54]. Neutralizing antibodies against AAV2 were observed to reduce the transduction efficiency [55]. Pre-existing antibodies to AAV8 were also shown to impact the biodistribution of the GTx [56]. Recent data from preclinical and clinical studies suggest that it may be possible to administer AAV5-gene therapy in the presence of anti-AAV5 antibodies below a certain threshold [57]. However, other reports demonstrate that anti-AAV antibodies above a certain threshold may limit successful gene therapy treatment [34,39,58], and that the level of transgene expression may be markedly lower in the presence of even low titer antibodies [59]. Hence, a cautionary approach which may require excluding patients positive with high titers of pre-existing antibodies is recommended for first-in-human (FIH) studies. However, it should be noted that once the approach of excluding patients based on their pre-existing anti-AAV Ab titer is implemented it will be very difficult to remove this exclusion criterion. Consideration should also be given to immunomodulation approaches using corticosteroids to mitigate the side effects of the administration of a GTx and IgG depletion approaches to reduce pre-existing anti-AAV Ab titers to increase the potential benefit of these therapies. To date, industry practice has been to use humoral pre-existing antibody responses for inclusion/exclusion criteria in trials (dependent upon the ROA). Therefore, it is important to understand the impact of pre-existing antibodies on clinical outcomes. Since a qualitatively similar immunogenicity response may have different in vivo responses, consensus was reached that a universal threshold of pre-existing antibody set as an inclusion/exclusion criteria is not relevant. The impact of other attributes of the pre-existing anti-AAV response on transduction, such as isotype, affinity, epitope recognition and potential to activate complement, are not yet understood. Sponsors are recommended to seek regulatory advice early to determine the appropriate assay to assess pre-existing antibodies with the understanding that the impact of pre-existing antibody is more nuanced than magnitude (titer) of response.
Creating strategies to enable the administration of GTx to seropositive patients with pre-existing antibodies would be valuable to a significant number of patients [60]. To re-administer GTx to seropositive subjects, immune modulation, adjustments to dosing and IgG removal were discussed as options [29]. One proposed approach to reduce or remove pre-existing antibodies is the use of transient enzymatic degradation, i.e., cleavage of intact human IgG by an IgG-degrading enzyme derived from Streptococcus pyogenes. This is a 2-step process and results in a fully cleaved IgG molecule that cannot mediate complement-dependent cytotoxicity (CDC) or antibody-dependent cellular cytotoxicity (ADCC) by means of the Fc receptor [61]. This approach was used successfully to treat and re-dose NHP with pre-existing antibodies to AAV8 [62]. Another option is the mechanical removal of pre-existing antibodies by plasmapheresis or immunoadsorption, also successfully demonstrated in NHP. Further characterization of the determinants of pre-existing immunity that may impact AAV transduction will allow us to better understand which patients will benefit from interventions.
When performing an immunogenicity risk assessment for GTx treatment, product and patient related factors should be taken into consideration. Product related factors that can be modified to reduce immune responses may include (but not be limited to) vector design, transgene optimization, critical quality attributes and the ROA. However, any such modifications should not have a negative impact on the risk:benefit analysis for the GTx. As already discussed, pre-existing antibodies, in particular NAbs, can significantly impact the efficacy of the treatment. Immunogenicity data from a prevalence study looking at pre-existing NAb and TAb to AAV2, AAV8 and AAV5 showed that approximately 50% of the study participants had pre-existing anti-AAV NAb with anti-AAV8 NAb positivity slightly lower than anti-AAV2 NAb or anti-AAV5 NAb [63]. After receiving an AAV-based gene therapy, the TAb and NAb titers increased markedly and remained positive for at least 2 years although the cellular responses against AAV8 as measured by Type II interferon enzyme-linked immune absorbent spot (IFN-γ ELISPOT) assay were variable among the recipients and showed no correlation with transgene expression [64,65]. The strong humoral response against AAV may prevent or reduce efficacy of future AAV-based gene therapy. It should be noted that the AAV8 based FVIII gene therapy in this study did not induce or boost humoral or cell mediated immune response to FVIII, thus did not impact the efficacy of FVIII replacement therapy where prophylactic or on-demand treatment with plasma derived or recombinant factor concentrates are the standard care for hemophilia patients.
For rare genetic diseases with replacement therapy as the standard of care, gene therapy provides hope for a cure. It is important to understand whether the severity of disease gene mutation and levels of residual gene product in individual subjects may influence the immune response to the transgene product. Clinical studies of biologic replacement therapy have shown that subjects with missense mutations had low Ab seroconversion incidences and generally better clinical outcomes than subjects with severe gene mutation (e.g., deletion, frame shift) partially due to replacement therapy products not having different epitopes therefore not being recognized as foreign [66]. Maintaining a threshold level of expression in systemic circulation may help with immune tolerance and reduce immune response to the transgene products. Another approach to reduce humoral and cellular immune response is via immunomodulation and immunosuppression, which helps to promote long-term expression of transgene products, potentially leading to improved clinical outcomes [64,65]. Hence, understanding the mechanisms behind immune response is key to designing an efficient immunogenicity mitigation strategy and monitoring for AAV-based GTx.
Commonly, detection of TAbs or NAbs have been utilized for patient enrollment decisions in gene therapy clinical trials. In fact, regulators confirmed that, more often than not, TAb assays are used to detect pre-existing antibodies for patient inclusion/exclusion. These methods have adequate sensitivity and robust performance over time. However, there is a lack of concordance in a small proportion of subjects who screen positive in the TI assay but do not have detectable TAb, indicating the potential need for a systematic study on the impact of pre-existing antibodies. In both clinical and non-clinical studies, the presence of neutralizing activity alone did not significantly impact the efficacy or safety of gene therapy administration. However, the impact of detectable pre-existing AAV-specific TAb, and the relationship of TAb titer and AAV vector dose level remains to be studied. In addition to humoral responses, the understanding of T cell cellular response to the vector and transgene product using an ELISpot assay may be relevant, as needed given the technical challenges. The timeframe for immune response monitoring (humoral or cellular) to assess safety and efficacy may be as follows:
- pre-treatment (i.e., screening stage) – for pre-existing anti-AAV antibodies
- immediately prior to administration – to determine whether seroconversion has occurred between screening and dose administration
- post-treatment – cellular and humoral responses to the vector and transgene product
Results in an NHP study using an AAV5-based gene therapy encoding human FVIII (hFVIII), demonstrated that positivity in the pre-existing antibodies detected via cell-based TI assay in the absence of AAV5-specific TAb, did not correlate to a lack of efficacy. Animals co-detected with neutralizing TI activity and AAV5 TAb showed poor transduction and reduced or absent expression of hFVIII [34]. Pre-existing immunity to hFVIII, also called FVIII inhibitors, may reduce efficacy by neutralizing the activity of the expressed protein and/or impacting therapeutic clearance. Patient safety is also a concern as these inhibitors have the potential to bind to the endogenous hFVIII in patients or the nascent protein expressed by the administered GTx. Hence, patients in the clinical trials were required to have no prior history of FVIII inhibitors and >150 exposure days to FVIII replacement products. Additionally, results from the NHP study informed the patient enrollment criteria in Phase III studies where subjects with detectable AAV5 TAb at screening were excluded. All subjects developed a sustained anti-AAV5 antibody response; as detected in the plate-based TAb assay and the cell-based TI assay; however, there was no difference in the pharmacodynamic (PD) readout (FVIII activity) between subjects that were positive or negative in the ELISpot assay used to assess cellular response. These responses declined or were negative in the long-term follow up period. The AAV5 capsid-specific cellular immune response may be a contributing factor to the transient increases in alanine transaminase responses in some patients.
In both clinical and non-clinical studies, the presence of neutralizing activity alone did not significantly impact the efficacy or safety of gene therapy administration. The impact of detectable pre-existing AAV-specific TAb, and the relationship of TAb titer and AAV vector dose level remains to be evaluated.
CRISPR/Cas9 Immunogenicity & Bioanalytical Challenges
Gene editing using clustered regularly interspaced short palindromic repeats (CRISPR)-Cas9/RNP is a relatively novel development in gene editing therapy. The therapy is a complex composed of single guide RNA (sgRNA) and Cas9 endonuclease (a bacterial protein) where the sgRNA “guides” the Cas9 to a specific DNA sequence in the cell and causes a specific break in the sequence. The promise of the therapy lies in the ability to exploit the cell’s DNA repair machinery by introducing changes to one or more genes.
Previous White Paper recommendations [25] focused on the risks in current ex vivo protocols. For example, the current protocols for ex vivo administration have a low risk of exposing the subject to the CRISPR-Cas9 machinery due to the long duration between expansion of the desired cell clone and subsequent administration, which allows for the RNP complex to degrade. Shorter ex vivo protocols increase the risk of a patient being exposed to the active/inactive RNP complex which could result in the potential for undesired in vivo editing and/or immune system initiation in the form of anti-Cas9 protein antibodies or T cell activation [67].
Understanding the complexity of the CRISPR-Cas9 gene editing therapy, mechanism of action, disease and product attributes will help in developing the appropriate bioanalytical strategy. A comprehensive assessment at the preclinical and clinical stages will help identify the intended use of the assays (i.e., biodistribution, activity of complex, off-target effects). This in turn will help with selecting the appropriate bioanalytical assay platform, developing and validating methods as needed. As discussed in the 2019 White Paper [25], the regulators’ current expectations for gene editing therapies are an understanding of the exposure and biodistribution data (total and active RNP complex) and, where feasible, the testing of tissue biopsies in clinical studies. In addition, screening subjects for pre-existing Cas9 protein specific immune responses (mainly antibodies) and post-treatment monitoring of these immune responses was recommended.
Sponsors are now developing approaches for the systemic non-viral delivery of CRISPR-Cas9 therapies as this mechanism has the potential to be curative with a single dose by gene editing, insertion or deletion. Non-viral delivery expands therapeutic utility of GTx by allowing for delivery to multiple tissue types [67–69]. Bioanalytical assays serve to facilitate the evaluation of safety and efficacy in addition to monitoring the newly expressed protein (when the therapy involves a gain in function) or the reduced protein expression (when it is a knock down of a gene). Case studies were discussed to demonstrate these uses. The first was to verify that the mechanism of action knocked out genes to remove mutant proteins; serum reduction of the protein (PD marker) in an NHP study was used to model first in human dose. These data were confirmed in the Phase I study [70]. In another study using NHPs where the mechanism of action was again to knock out genes, monitoring of the reduction in protein levels and activity in NHP was used to predict human dose [71]. In addition to gene knock out, in vivo administered CRISPR-Cas9 can be used for inserting genes to restore normal protein function. A NHP study using measure of the restored protein was useful in designing the clinical study [72].
Discussions at the 15th WRIB focussed on understanding some of the risks of the off-target toxicity between ex vivo and in vivo administered CRISPR/Cas9 therapies. There is current insufficient data to draft guidance, but it was recommended to generate data to support patient safety by the use of in vitro assessments [73] as well as the use of animal models while acknowledging that these may not always translate in the clinic. These studies could include the use of T cell receptors on CD-4 cells and studying protein engagement using major histocompatibility complex (MHC) Class I molecule presenting cells.
With regards to immunogenicity, while there are several publications on pre-existing anti-Cas9 protein antibodies due to natural environmental exposure to streptococcus and staphylococcus [67,69], additional discussion is needed. The potential impact of anti-Cas9 protein antibodies to pharmacokinetics (PK), PD, and safety is unknown at this time. In addition, the observed antibody to Cas9 protein is not expected to cross-react with Cas9 mRNA as a component of the drug. It was also acknowledged that pre-screening for Cas9 protein antibodies may not be feasible for orphan or rare disease populations.
CAR-T Immunogenicity & Cellular Kinetics
General Considerations
The delivery of CAR-T cell therapies presents unprecedented opportunities and challenges for bioanalytical scientists. Persistence of CAR-T cells in circulation plays a critical role in long term efficacy and can also pose a potential safety risk during treatment and after remission. Due to the unique design of CAR-T cell therapies, both humoral and cellular (cytotoxic T cell-derived) responses would contribute to CAR-T immunogenicity. There is a regulatory expectation that these responses are monitored to help in the characterization and de-risking of a new CAR-T therapy [41,74].
Like other biologics, immunogenicity assessment methods implemented at different stages of drug development are based on the immunogenicity risk assessment. As autologous T cells present a low to medium immunogenicity risk, an anti-CAR antibody assay is typically implemented at the start of the FIH study. Anti-CAR antibodies may be detected using classic LBAs. For NAb assessment, an assay platform that reflects the drug mechanism of action is expected, and assays should be designed such that the end point reflects the inhibition of the biological function of the drug. For CAR-T products, NAb assays are challenging and there has not been specifically designed NAb assay schemes published. In addition to the competitive ligand binding (CLB) assays, a cytotoxicity assay may also be used for cellular immunogenicity detection. The typical format for a cytotoxicity assay involves co-culturing of CAR-T and target cells. This method does have some challenges; it may be difficult to obtain patient CAR-T cells and the human leukocyte antigen (HLA) types between CAR-T and target cells may have to be matched to eliminate background killing. The use of engineered surrogate CAR-T cells or target cells carrying the reporter gene may overcome some of these issues. Due to the technical challenges associated with a cytotoxicity assay, the CLB assay is considered a preferred assay format that is usually more sensitive with reliable assay performance.
Humoral Immunogenicity
For complex molecules such as CAR-T which have many immunogenic domains, and the potential to elicit a host of immune responses (e.g., anti-scFv antibody, anti-hinge, anti-linker antibody), a bridging LBA with soluble CAR may be adequate. In contrast, an ELISA bridging source antibodies may only detect immunogenicity against epitopes that are part of the scFv domain of the CAR. To detect the other aspects of the immune response, such as anti-membrane protein interaction epitopes and anti-insoluble extracellular domains, a cell-based method using flow cytometry may be more valuable [75]. Given that the vast majority of CAR-T therapies are in clinical development [76], information on the immunogenicity of CAR-T products is quite limited; the consequences of immunogenicity for this class of therapeutics are largely unknown although they are being monitored in the on-going clinical studies. The use of these different assay techniques offers insights into different aspects of immunogenicity, with challenges for sample collection and validation strategy for cellular response assessment and new opportunities that allow immunogenicity risk assessment and mapping of the ADA response for this novel treatment modality. Industry requests product specific guidance from the regulators.
There are limitations in the use of LBAs for CAR-T therapies. CAR domains used may not represent the full extracellular domain (ECD) in vivo. There is the risk of masking immunogenic epitopes due to the labelling required for these formats. In addition, conformational epitopes [77,78] and tertiary structural differences may be missed. A flow cytometry-based assay using recombinant cell line expressing CAR or a cell membrane associated ECD domain immobilized on plate format can be an alternative for anti-CAR antibody detection as previously mentioned.
For autologous CAR-T therapies, host humoral immune responses can be measured with LBA or cell-based assay formats [75,79]. Regulators noted that screening assays are commonly based on the CAR ECD, the source antibody of the CAR scFv domain, or the scFv domain only, whereas intracellular CAR domains are not typically included. As with other immunogenicity assays, matrix interference must be mitigated. In addition, parameters like drug tolerance, sensitivity, and precision should be evaluated [52,53] to ensure that the method is fit for purpose. Approaches to improve drug tolerance such as affinity capture elution (ACE) [80] may also help with reducing matrix interference.
Autologous bi-paratonic CAR therapy increases the avidity of traditional CAR; two scFv domains recognize different epitopes of the same tumor antigen. This presents some unique challenges in the assessment of NAb response. Like other multi-domain biologics, it is important to assess immune responses to the CAR ECD as well as the individual domains. Multiplex platforms allow for the simultaneous assessment of NAb to the full-length CAR and two separate scFv binding domains, respectively.
The advent of CAR-T therapies has given rise to novel uses for existing platforms such as flow cytometry, e.g., for the detection of the binding of ADA on CAR-T cells. Previous recommendations stated that the methods should be FFP, and there was clear consensus that assay controls are to be implemented to ensure consistent assay performance [25,28]. However, critical reagents like CAR expressing cells or PCs, may be challenging to procure for the development of flow cytometry assays. For example, for autologous CAR-T, while patient-specific T cells are ideal controls, their use is often prioritized for drug product lot release, resulting in limited availability for use in method development and as assay controls. Furthermore, staining a patient’s blood or PBMCs to identify the CAR transduced T cells may interfere with actual ADA detection. An option to overcome these challenges is the use of the CAR transduced cell lines that offer high transduction rates and consistent CAR expression. Antibodies binding to these transduced cells can then be detected with anti-human IgG/IgM F(ab’)2 or anti-human IgG Fc [75,81,82]. This assay format has the advantage of consistent performance and CAR expression. However, the challenge with using this approach is that unrelated receptors on the selected cell line can give rise to the nonspecific signal in the assay. The use of wild type T cells may be needed as a control to identify (and possibly exclude) non-CAR specific responses [75].
There are some advantages to using flow cytometry. The CAR proteins do not have to be labelled, minimizing the risk of masked epitopes. Flow cytometry is not limited by the presence of insoluble/hydrophobic domains and CARs processed through the T cell expression system (e.g., glycosylation), allowing for CAR presentation in the native environment and conformation along with potential interaction partners. Hence, it is thought that antibodies detected in this format may be more clinically relevant.
There are some unique method development challenges, however, in terms of anti-CAR antibody staining and gating strategies to be used. Cell populations may ‘shift’ over time in the FSC/SSC (forward scatter/side scatter) plot, necessitating the setting of appropriate gates and performing daily QC performance checks of the cytometer. Another challenge is determining how to set an appropriate assay cut point. There may be a high prevalence of anti-CAR antibodies resulting in the inability to distinguish distinct positive and negative populations. This is not unexpected given that CAR-T therapies frequently contain mouse single-chain variable fragments and fusion sites in the chimera [81]. There is abundant evidence that human anti-animal antibodies, most commonly human anti-mouse antibodies (HAMAs), are present in up to 80% of humans [83], and subjects who received CAR-T infusion were seropositive for HAMAs post infusion [84]. Hence, minimizing the inclusion of mouse components when designing CAR-T therapies may be important [77]. If there is high prevalence of anti-CAR antibodies [75], assay controls could be prepared in immunoglobulin depleted serum. However, this matrix for controls is different from study serum samples and can result in extremely low cut points. A more common mitigation approach is to consider patient specific cut points – using the patient pre-treatment samples to calculate the cut point. Statistical approaches should also be considered [85]. Regulators are not prescriptive on the methodology to set the assay cut point but are looking for justification of the approach used.
Cellular Immunogenicity Assays
The evaluation of cellular immunogenicity using the expression of cytokines as markers of activation (e.g., IFN-γ on CD8+ cytotoxic T cells and CD4+ helper cells) has potential uses in investigating the mechanisms of CAR immunogenicity. This in turn may help with understanding the associated immunogenicity risk. Cellular immunogenicity plays an increased role in immunogenicity particularly if B cells are depleted in the patient population. Two commonly used platforms for the assessment of cellular immunogenicity are ELISpot and FluoroSpot; the underlying principle is to detect cytokine secretion (e.g., IFN-γ) of activated T cells as a functional measure of antigen-specific cellular immune response to CAR. To date, there has been a low incidence of cellular immune responses to CAR-T therapies and a clear relationship between cellular immunity and clinical outcome has yet to be established [28].
The cellular immune response against CAR can be assessed using a patient’s PBMCs. After stimulation of PBMCs with CAR peptides, activated T cells excrete IFN-γ that can bind to coated anti-IFN-γ antibodies. After removal of cells, spots with bound IFN-γ are detected with either peroxidase or fluorophore-labeled secondary antibody. The spots can then be quantified using colorimetric substrate or fluorescence analysis. As illustrated in the case study discussed, cellular response to CAR-T administration was evaluated, where patient PBMC samples were stimulated using 4 peptide pools consisting of 15-mer peptides. Assay medium was used as the negative control and phytohemagglutinin (PHA) was used as the PC. In addition, PBMC controls were selected from a library of donor controls such that post PHA activation, the cells would produce <10, ∼300 and >1000 spots. Patient samples that showed ≥3x increase in spot number in post-treatment when compared to pre-treatment and ≥3x spot number of media control were considered positive. Assessment of samples showed a higher frequency of cellular immunogenicity directed against scFv peptides. While there was a trend towards lower levels of CAR-T cells in patients with anti-CAR T cell response, this was not statistically significant one month post infusion [86].
Cellular immunogenicity can also be analyzed using a 51Chromium release assay. After stimulation with autologous, irradiated CAR-T cells and cytokines (IL-12), the potential of a patient’s PBMCs to lyse 51Chromium-labeled CAR transduced T cells is quantified. For a response to be counted as positive, PBMCs must lyse CAR-T cells but not transduced T cells. To illustrate, a case study was discussed where post-infusion lysis of CAR-T cells was described for six out of eleven patients [87].
Intracellular cytokine staining (ICS) can also be used as a functional measure of antigen-specific cellular immune response to CAR. The method involves the addition of CAR or CAR peptides to patient PBMCs, followed by the fixation, permeabilization and staining with multiple dyes to identify the relevant cell populations and includes proper controls to minimize false positive results. On a flow cytometer, the percentage of T cells expressing IFN-γ can then be determined. This format has the benefit that, depending on the staining applied, it can distinguish between CD4+ and CD8+ T cells [88].
There are some general considerations and a few practical tips in the development of flow cytometric immunogenicity assays. The use of sodium azide and low temperatures can prevent internalization of surface antigens; Fc receptors can be blocked using specific reagents such as Fc block; complement activity can be reduced by heat inactivation. In terms of secondary antibody, the selected antibody should not bind the scFv portion of the CAR directly. The use of Fab(2) detection antibodies may help mitigate the potential to bind Fc receptors.
In summary, overall T cell functional assays have higher variability. Quality data is critically dependent on cell cryopreservation for cell viability and functionality, as well as sufficient cell numbers (and total blood volume). These challenges need to be considered prior to implement such tests as part of clinical studies.
Bioanalysis of Cellular Kinetics
One of the key challenges facing CAR-T therapies is the development of a more complete understanding of the exposure-effect relationship. For CAR-T therapeutics additional variables need to be taken into consideration; tumor antigen expression levels, CAR expression, rate of expansion/contraction/persistence of CAR-T, and tissue distribution are all factors that need to be considered. The elucidation of this relationship starts with the development of a robust bioanalytical strategy. For CAR-T therapies, direct and indirect approaches can be used to measure cellular kinetics. Flow cytometry assays quantify CAR-T cells directly and can also provide information on the distribution of CAR expression. Alternatively, qPCR and ddPCR assays that quantify the copy numbers of transgene can serve as an indirect measure [89–91]. Each bioanalytical method has distinct advantages and disadvantages for the quantification of CAR-T therapies (Table 2); both technical and operational differences need to be considered when selecting a method. For example, the absence of a normalization factor coupled with the relative instability of CAR-T cells necessitates that samples be analyzed in real time for flow cytometry assays which causes logistical challenges as well as increased costs. On the other hand, samples can be banked for transgene quantification by qPCR or ddPCR and run in a batch format; however, transgene measurement is a surrogate readout for CAR-T cells. Both of these categories of differences need to be considered in the context of the question that is being asked.
| Attribute | qPCR/ddPCR | Flow Cytometry |
|---|---|---|
| Regulatory guidelines | No bioanalytical guidelines for PK assessment | No bioanalytical guidelines for PK assessment |
| Analyte | Genomic DNA | Cell surface protein |
| Sample preparation | Well established DNA extraction techniques from numerous tissues | Requires single-cell suspension of PBMCs |
| Normalization | Copies normalized to DNA input through a reference gene | No Normalization |
| Sample requirements | 100 to 400 ng/reaction of DNA | 50–100 μl blood; 250,000–1 million cells |
| Sensitivity | ≤50 copies/μg; 50 copies/149,925 cells (0.033%) | Dependent on multiple variables |
| Stability | Isolated DNA is stable | Samples must be analyzed shortly after collection for maximum sensitivity |
| Logistical considerations | Samples can be frozen at clinical sight and stored prior to DNA extraction and subsequent bioanalysis | Samples must be analyzed fresh for maximum sensitivity because of limited stability |
| Analysis | Straight-forward (interpolation off a standard curve) | Requires trained analyst to use the instrument, perform gating and analysis |
| Key reagents | Primers and probes: can be synthesized for any sequence | Antibodies against cell surface antigens: may require significant time for generation |
| Multi-signal analysis | Yes | Yes |
| Units | Copies/μg DNA | Cells/μl blood; enumeration of cell type |
| Assay interference | DNA-binders (i.e., heparin); not susceptible to interference from ADA | Potentially susceptible to ADAs or soluble target proteins |
Ultimately, it is important to question how the exposure-effect relationship differs between flow cytometry and qPCR/ddPCR (CAR transgene) derived bioanalytical data. To address this, two data sets from a Breyanzi clinical study were compared to each other, one determined from a flow cytometry assay and the other from qPCR-based assay [92]. Breyanzi is CD-19 targeting CAR-T that is approved for the treatment of relapsed-refractory B cell lymphoma. There was a strong correlation between these two data sets and the PK parameters (Cmax and AUC) were comparable to each other, thus the exposure-effect relationship is expected to be similar between these two assays.
Both flow cytometry and qPCR/ddPCR can be used to understand cellular kinetics. The choice of method should reflect the consideration of the research question at hand as well as the advantages and disadvantages of each method.
Characterizing the in vivo cellular kinetics (CK) and biodistribution of CAR-T cells is critical for preclinical studies, clinical toxicity and efficacy assessments of CAR-T cell therapies. However, there is no standardized assay to characterize CAR-T cell distribution, expansion, contraction, and persistence profiles. To overcome this limitation and facilitate data comparison across different studies it was recommended to develop a universal duplex ddPCR assay protocol for CAR-T CK and biodistribution analysis. This duplex ddPCR assay should utilize a single copy gene as the reference gene (RG) to normalize the genomic DNA (gDNA) input. To enable the use of the same assay across species, the primer/probe for the reference gene should be designed to target conserved DNA sequences in the genome (e.g., between mouse and human). When combined with white blood cell count results, the duplex ddPCR assay raw data for blood CK samples can be readily expressed as copies/μg gDNA, copies/μl blood, or copies/diploid cell to meet the current regulatory requirements and allow for more accurate and systematic evaluation of CAR-T cell expansion and a direct comparison with flow cytometry data.
qPCR/ddPCR Assay Performance
There has been exponential growth in the use of qPCR and ddPCR technologies matching the growth in gene therapy, cell therapy and vaccines. The most commonly used platform to determine biodistribution has been qPCR, and transgene expression is evaluated by RT-qPCR or by the measurement of the expressed protein using LBA or liquid chromatography mass spectrometry (LCMS) platforms [21]. After a further year of experience applying the 2019 and 2020 White Paper in Bioanalysis recommendations, the 2021 discussions focused on the use of the appropriate technology platforms, validation guidelines and unanswered questions [25,28].
qPCR
Discussions focused on using RT-qPCR to quantify transgene copy numbers and relative expression of target genes as target engagement (TE) biomarkers in gene therapy. Genome regulation by modulating the expression of disease-related genes is a promising application of gene therapy. This is done by the in vivo delivery of a transgene encoding a transcriptional regulator that can selectively bind to a region of genomic DNA and up- or down-regulate the expression of a targeted gene to varying extents. Hence, target gene and transgene expression can be used as PD markers if they can be measured in vivo. In early pre-clinical lead selection and IND-enabling studies, when establishing TE in the tissues of interest, it is important to establish the correlation between the absolute quantitation of the transgene (mRNA copy number/cell in treated samples) and the fold change of target gene expression by comparing treated vs untreated samples. This is done using RT-qPCR. While the RT-qPCR methodology is fairly standard, there are questions about generating reagents to use for standard (DNA or RNA) preparation and the variability of the efficiency of the RT reaction.
One of the parameters that introduces variability in the RT-qPCR methods is the efficiency of the RT step, potentially impacted by the mRNA encoded by different genes, mRNA integrity and purity, the enzyme efficiency and primer selection strategy. To determine RNA integrity and purity, the RNA integrity number (RIN) can be used as a measure. Using RNA standards enables absolute quantitation by accounting for RT efficiency and variability which enables a comparison of transcript levels across studies.
Two approaches for generating RNA reference materials were discussed; the in vitro transcribed method or synthetic RNA. Given the general poor stability of RNA that are subject to exo- and/or endo-nuclease degradation, an alternate method could be to use double stranded DNA (dsDNA) as standard material as these are easy to implement and will save time, however, the RT efficiency and variability will not be accounted for, and this method will not be considered absolute quantitation. For QC selection/preparation, spiked RNA can be used, however this approach does not reflect the RNA recovery from the homogenization/cell lysis process and the degree of degradation of target RNA during this process cannot be accurately measured.
When absolute quantification methods have been used to detect transgene mRNA copy numbers, the standards and controls were prepared in matrix RNA from relevant species and tissues. The assay parameters selected for evaluation are based on those routinely used for LBAs [93,94]; range of quantitation, sensitivity (copy/reaction), selectivity (at least 8/10 individual matrix RNA spiked with QC should recover within 80–120%), precision and accuracy, and stability.
For relative quantitation of gene expression using the ΔΔCt, method, several considerations for method development were discussed. RGs are recommended to control for sample input and should be unaffected by treatment and constantly expressed with a high amplification efficiency. In terms of the primer selection, there should be equal amplification efficiency (≤5%) between the target gene and RG gene(s). To monitor run to run performance of these methods, like other assays, QCs are needed. However, the classic spike-in of the reference material cannot be considered for relative quantitation. Instead, one can prepare the QCs by diluting the total RNA isolated from relevant naive species/tissues in carrier RNA and determining precision in multiple runs (e.g., n = 5). The %coefficient of variation (CV) should be ≤2%Ct and, when normalized to RG genes, should recover within ±25% of the undiluted QC. An alternate approach to consider is the use of treated and untreated samples from prior studies (if available).
Discussions revealed that the absolute quantification method is not often used for the broader gene therapy applications, since quantification may not be needed for the ultimate use of the data (e.g., demonstration of presence of the transgene in the tissue of interest) and in the clinic, access to tissue samples may not be possible. Furthermore, there is currently insufficient data to conclude that mRNA measurements are true surrogates for the protein of interest. If absolute quantification is used, mRNA should be reported as copy/μg of total RNA and not normalized to mass of tissue.
Characterization of viral vector shedding is critical to minimize the risk of disseminating infectious material [41,42] and qPCR or ddPCR is the methods of choice to determine the total shed vector. To guide industry on the pre-clinical and clinical development, health agencies have published several regulatory guidelines for gene therapy development including some recommendations and considerations related to bioanalysis [41–46,95,96]. The most recently published draft guidance for industry is on human gene therapy for neurodegenerative diseases [97]. In addition, the Global Bioanalytical Community has leveraged the MIQE guidance [50] along with several White Papers that address assay considerations [21,25,28]. A fit for purpose validation approach using standards, QC and no template control (NTC) is recommended, with calibrator material equivalent or very similar to test samples. If the assay is intended to be used over an extended period of time, it is preferred to use current Good Manufacturing Practices (cGMP) material, however, this may not be feasible particularly in the early stages of development. If research grade material is used, bridging to the lot of clinical material should be considered as soon as possible.
During method development for qPCR assays, in addition to the analyte considerations, aspects such as primer and probe design, DNA extraction/optimization for each tissue/biofluid of interest as well as assay range, sensitivity and linearity using a linearized vector as standard should be carefully considered. Selection of primer and probe is key and should be evaluated with an eye toward the long-term maintenance of the assay (assay life cycle). As such, it was recommended to avoid primer dimers and make selections based on physical properties, amplicon length, efficiency (90–110%, 85–90% may be acceptable for early studies), sensitivity and specificity. Ideally, the same primer/probe pair that can be used across pre-clinical and clinical studies should be selected. For biodistribution and shedding studies, given the variety of tissues/biofluids involved particularly in the pre-clinical phase, it is important to identify primary and secondary tissues/biofluids of interest and to focus assay development and validation on the primary matrices. To get optimal results, it is important to start with as high a recovery as possible and hence DNA extraction experiments should be optimized to have recovery in the range of 30–80% with >50% for primary tissue/biofluids. Table 3 summarizes the updates or refinements to consensus in qPCR method development practices compared to prior recommendations [25,28].
| Validation Parameter | 2019/2020 White Paper Recommendations | 2021 Recommendations Based on New Applications in qPCR method development/validation | Status of Consensus |
|---|---|---|---|
| Standard Curve – Dynamic Range | • Precision: Ct values (threshold cycle) • Accuracy: back-calculated copy numbers | • Precision: %CV of Ct between replicates • Accuracy: %RE of back-calculated copies • Inter-assay reproducibility: %CV of Ct at each level • 75% and a minimum of 6 non-zero standards should meet acceptance criteria | Refinement of the consensus (especially in terms of number of standards needed) |
| Standard Curve – Linearity | Not explicitly addressed | • R2 >0.98 • Slope between -3.58 and -3.10 corresponding to 90–110% PCR efficiency | Refinement of the consensus |
| LLOQ and LOD | • Transgene: ≤50† copies/μg gDNA (149,925 cells, 0.03%) • Reference gene: can be much higher (<300,000 copies/μg gDNA) | • LLOQ: Ct<40 • LOD: at least 95% of samples are positive i.e., Ct < Ct of NTC and Ct<40 • LLOQ ≤50 vector copy numbers per μg DNA if applicable† | Consensus with some refinement |
| QC | • 3 levels • 3 replicates per sample: duplicate reactions plus one replicate spiked with qualified internal control (no need if multiplexing) or multiplexed into each sample | • 3 to 5 levels spiked into gDNA • at least 2 replicates and 3 replicates of NTC • Acceptance criteria are 2/3 of QC and ≥50% at each level pass with NTC<LOD | Consensus with some refinement |
| Specificity | Need to distinguish sequence of gene of interest sequence from other interfering endogenous sequence (check for cross reactivity) | No specific amplification (<LOD) of target DNA by primer/probe combination when spiked in non-specific related target into matrix | Consensus with some refinement |
| Selectivity | Not explicitly addressed | • At least 10 individual gDNA samples or tissue lysates • Tested unspiked and spiked with target DNA • At least 80% of spiked samples should have acceptable accuracy and precision | Refinement as knowledge builds |
| Stability | Storage for each matrix and DNA using spike in controls and processing | • Bench-top and freeze/thaw stability of QCs (at least low and high QC levels) • %RE of copies relative to freshly prepared QCs | Consensus with some refinement |
| Incurred Sample Reanalysis (ISR) | Limited data available for ISR to outline suitable criteria | Not considered | Divergence or not enough data at this time |
†The assay should have a demonstrated limit of quantitation of <50 copies/μg genomic DNA, so that your assay can detect this limit with 95% confidence [44].
qPCR in vaccine development
Traditionally, qPCR assays have played a pivotal role in the primary endpoint analysis of vaccine studies (short- and long-term vaccine efficacy). Previous White Paper recommendations [25,28] have provided guidance in terms of assay development considerations and parameters for validation, but only in the context of non-vaccine modalities (i.e., CAR-T and gene therapies). For vaccine development, particularly when using molecular methods for non-diagnostic uses, it is important to share best practices and recommendations with a particular focus on LOD, LLOQ and False Positive Probability (the probability of obtaining a positive result given a negative sample). Parameters used for vaccine serology assays [28] are still relevant (precision & accuracy, specificity, robustness & ruggedness), however LOD, limits of quantification and false positive probability are particularly important in the context of using these methodologies for vaccine development [98–101].
The LOD, defined as the lowest amount of target organism detected but not necessarily quantified, is evaluated using a series of dilutions of the specific organism in negative samples. The percent positive results (# positives/total*100) are plotted against input copy number and fit with an appropriate regression model. The LOD is the copy number at which the model predicts the assay can detect the specific organism 95% of the time.
For qPCR assays, it is important to establish the limits of quantification. To do so, a standard curve is prepared by the serial dilution of the selected reference standard. In the example discussed, a series of dilutions of a viral pathogen was prepared in discrete negative samples and RNA was extracted by at least two analysts in replicate. The extracted RNA was evaluated in the RT-qPCR assay and a geometric median concentration and variability estimate was calculated. The reference standard used in this method is critical as not only the LOQs but also the LOD are only as good as the copy number assigned to the reference standard. It is preferred and recommended to use a WHO reference standard (if available) to calibrate lab-derived reference standards.
A myriad of approaches exists to generate reference standards: high copy human sample, engineered constructs (plasmid), encapsulated nucleic acid, and purified pathogen from an in vitro culture [102]. In the absence of an international reference standard, it can be challenging to assign a copy number to the reference standard; two approaches to consider, each with their own limitations, are mass calculation and digital PCR (dPCR) using the same primer/probe set used in the quantitative assay. Mass calculation-generated standards could include non-amplifiable copies or host cell nucleic acid contamination while those generated using dPCR may be impacted by reporter dye compatibility and the multi-log dilution needed to quantify the standard.
Because PCR is very sensitive and prone to contamination, the use of engineering and procedural controls from purification through PCR analysis should be considered to minimize the risk of contamination. The determination of false positive probability is an essential part of the validation and should be used to assess the contamination controls. This is accomplished by assessing a large number (>1000) of confirmed negative samples interspersed with positive samples. Both positive and negative samples are purified and processed in an identical manner and the number of positive and negative samples tallied. A binomial distribution is used to calculate the upper bound on a 95% confidence interval (if the numbers of positive and negative samples are small, a point estimate can be considered).
ddPCR
ddPCR is an emerging technology with some advantages over conventional qPCR, such as the ability to multiplex, the improved tolerance to variable efficiency of amplification, no need for a standard curve during sample analysis, and less matrix interference (refer to Table 4). ddPCR also allows to develop assays that are impossible to be performed using qPCR such as determining the linkage between two different sites on a vector as a measure of vector integrity [28]. The 2020 White Paper discussed the use of ddPCR to evaluate biodistribution and shedding of the transgene and the need for guidance on the use of this technology [28].
| qPCR | ddPCR | |
|---|---|---|
| External standard curve for absolute quantification | Yes | No |
| Reliance on amplification efficiency for quantitation | High | Less |
| Tolerance to PCR inhibitor | Poor | Excellent |
| Precision | Good | Excellent |
| Sensitivity/rare target quantification | Good | Excellent |
| Robustness | Excellent | Excellent |
| Multiplexing | Easy | Very Easy |
| Target quantification using DNA purified from various sample types with diverse quality/purity (biodistribution) | Very challenging | Easy |
| Dynamic range | Wide | Narrow |
| Dilution linearity | Good | Excellent |
| Validation | Standard methodology (see Table 3) | Not yet established |
To ensure that the assay is fit for its intended purpose, and given that there is currently no guidance for such assays, it is recommended to combine aspects of the BAV recommendations [103] and the long-term follow up post administration of GTx products [44]. QC samples can be prepared by spiking plasmid containing a species-specific CAR target sequence in mouse gDNA (pre-clinical studies) or human gDNA (clinical studies). For each 96-well plate run, at least two sets of QCs should be included and tested in triplicate. While the traditional LBA parameters of assay reliability are recommended, such as precision and accuracy, sensitivity, selectivity, LLOQ, and short-term stability (freeze/thaw, refrigerated stability), the acceptance criteria for precision and accuracy are vastly different. Table 5 summarizes suggested acceptance criteria.
| Parameter | Assay Acceptance Criteria |
|---|---|
| LOD | Lowest copy number that the statistical model predicts the assay can detect 95% of the time and higher than the observed contamination level (mean copy number detected in blank samples or NTC + 3.3 fold standard deviation) |
| LLOQ | Lowest concentration of an analyte in a sample that can be quantitatively determined with suitable accuracy and precision |
| Accuracy and Precision | ±35% bias and ≤40% CV for samples with ≥50 copies/20 μL ddPCR reaction ±50% bias and ≤80% CV for samples with <50 copies/20 μL ddPCR reaction |
Approaches for determining the nominal value of the prepared QCs was also discussed. It was recommended to use the mean of at least 3 repeat measurements, where each measurement is performed in triplicate. Table 6 recommends plate acceptance criteria.
| Parameter | Acceptance Criteria |
|---|---|
| Number of droplets/well | ≥10000 acceptable droplets/well |
| NTC | <2 detectable copies for CAR-T and reference gene |
| High and mid QC levels | ±35% bias and ≤40% CV for the measured CAR-T and reference gene copy number at each level |
| Low QC levels | ±50% bias and ≤80% CV for the measured CAR-T and ±35% bias and ≤40% CV for the measured reference gene copy number |
User experience with the ddPCR platform indicates that the technology is reliable, robust, and suitable to precisely and accurately quantify CAR-T in preclinical and clinical CK and biodistribution samples. Since formal regulatory guidance does not exist, it was recommended to provide a rationale for the approach to validation.
Analyte Stability in Vaccine Serology Assays
Vaccine serology assays (immunoassays) used to assess immune responses to vaccines form the basis for approval. They can also serve as correlates of protection when shown to be predictors of clinical benefit. It may be beneficial to maintain these assays in a validated state to support post marketing commitments and changes such as manufacturing process changes and new age indications.
For ligand binding and functional assays, the purpose of conducting sample stability studies is to ensure that the clinical test results are not influenced by sample handling (e.g., freeze-thaw) and/or storage conditions. Analyte stability is assessed under defined sample storage conditions and should optimally mimic the conditions under which clinical samples are collected, processed and stored. Additional stability studies may be needed depending on observed storage conditions (e.g., temperature excursions).
For protein biotherapeutics, the method validation guidance for pharmacokinetic (PK) assays lays out detailed procedural requirements for assessing analyte stability using low and high QC samples [93,94,104], but for ADA method validation, industry White Paper recommendations with regards to stability testing have evolved. Early publications [105] recommend the use of high and low levels of antibody as assay controls and as surrogates to assess sample stability. Shankar et al. recommended the use of incurred samples where possible as positive controls may not be reflective of study samples, and considers that stability assessments may not be needed when samples are stored at or below -60°C [106]. FDA acknowledges that while it may be difficult to establish sample stability, the data from such assessments may be useful. The guidance recommends storing samples to preserve antibody reactivity [52].
It is important to consider differences between ADA and vaccine assays; immunogenicity assessments for biotherapeutics are focused on the detection of small amounts of ADA (sensitivity) and mostly rely on the use of positive controls, often sera from hyperimmunized animals. This has limitations, as it is well understood that these controls do not reflect the immune responses found in clinical study samples [52]. Assays to detect immune responses to vaccines focus on the reproducible quantitation of antibody response to the vaccine and use incurred samples (from vaccinated or infected subjects) that spans the assay quantitation range. If incurred samples are not available, pooled samples or spiked samples may be used. Unlike ADA assays, there are no guidance documents for vaccine assay development, validation and maintenance, hence, industry has relied on the sharing of best practices in White Papers in Bioanalysis [25,28]. Specifically, the 2020 White Paper in Bioanalysis [28] provided recommendations on parameters to consider for serology assays; one of which is analyte stability.
Given the long duration of vaccine clinical trials, understanding long-term analyte stability is critical. However, as the variable moiety of antibodies has very little (if any) impact on their stability, repeating the experiment for every single assay may not be considered necessary given that there is published evidence documenting stability of antibodies (including IgG and IgM isotypes) in frozen serum at less than -60°C for up to 4–5 years [106–109]. In addition, the US Pharmacopeia states that serum samples stored at less than -70°C are stable and long-term stability is not required [110].
An alternate or supporting approach to assess long-term sample stability could be the use of proficiency panel data. A proficiency panel is a collection of samples that are routinely tested and used to monitor assay performance over long periods of time. These panels contain incurred samples that are evaluated on a regular basis using a pre-specified plan. With this approach, samples spanning the quantitative range of the assay are selected and evaluated periodically to set acceptance criteria [111]. Proficiency panel testing is set up in a phased manner: (i) A sample panel is identified in the screening phase; (ii) Baseline criteria are established; (iii) Acceptance criteria are set; (iv) The panel is used routinely. Samples to be used in the proficiency panel are selected on the basis of sufficient volume and desired titer that typically span the dynamic range of the assay. Baseline criteria are generated by the repeated testing of samples over a short time frame to create potential acceptance criteria which are confirmed/adapted after a longer period of testing to account for the natural and normal variation of the assay. The panel is then used for routine testing with a priori acceptance criteria. Given that the panel uses incurred samples, these data can be leveraged to also confirm or extend sample stability.
In the case studies discussed, retrospective analysis determined long-term analyte stability at -20°C and -80°C. However, real-time assessment of stability (e.g., using incurred samples from the study in question) would not enable assessment of the needed stability in a reasonable time and therefore re-testing data from incurred samples from a previous study was used to establish stability.
In addition to the assessment of stability, trending QC performance data was discussed. The goal of QC trending is to investigate and track the robustness of the method. It was recommended to use tools for tracking and trending, such as an exponentially weighted moving average [112]. In addition, method parameters that may not necessarily have acceptance criteria (e.g., cell viability, slope, functional output) can also be used to monitor trends in method performance and drift.
Vaccine Critical Reagent Management & Bridging
Critical reagents are the essential building blocks for high performing immunoassays and the need to generate and characterize reagents early in development is not new. A sustainable and reliable supply of reagents along with a replacement plan is essential to maintain acceptable assay performance over the life of the drug program. The use of multiple reagent lots over the life of the program is inevitable and minor differences in the biophysical characteristics of these reagents can impact assay performance. To minimize risk of interruptions and delays to the study, it is important to take a LCM approach to critical reagents [25,27], which includes producing large lots of reagents, characterizing them and preferably storing them in single use aliquots. This, however, comes with the risk of tracking stability of the reagent over the life of the assay. There are several White Papers that provide recommendations for expiry ranges for monoclonal and polyclonal antibodies (labeled or unlabeled), recombinant proteins and Fc fusion proteins [113,114]. These recommendations have stood the bioanalytical community in good stead.
A new critical reagent lot does not always increase method variability but can introduce a bias that can impact study conclusions despite large study population variability. Hence, the conduct of critical reagent bridging prior to use in clinical sample testing is key [113]. When designing appropriate bridging experiments and defining acceptance criteria, it was recommended to understand the intended use of the assay (e.g., considering the population variability, clinical trial protocol and statistical analysis plan) and assay performance (i.e., stability monitoring data, previous bridging data).
When bridging lots, critical reagents should first be evaluated for their biophysical and chemical properties. The use of label-free biosensor technologies such as Biacore™, Octet®, or Carterra® to determine specificity, epitope binning and testing of antibody pairs has previously been discussed [107]; this can be followed by evaluation of assay performance parameters using current and new reagent lots in a head-to-head comparison. In addition, making use of trending to observe reagent dependent assay parameters is recommended. Although it is ideal to evaluate a panel of incurred samples in multiple independent assessments with the candidate and qualified reagent lots, minimally, the performance of QC samples should be assessed. If a qualified lot of reagent is not available for comparison purposes (e.g., expired), a comparison to historically generated data may be considered [28]. The intended use of the assay allows for the setting of the maximum allowable bias or the real difference that can be tolerated between lots. Assay performance, and specifically assay variability, drives the detection of noticeable differences. Hence, for each assay and reagent type, it is important to have a prospectively defined and detailed bridging plan that includes acceptability criteria with an optimal sample panel size range (including a minimum number with valid final results), the distribution of samples (including a range of responses), the appropriate statistical parameters to be analyzed, and the potential impact of accepting a reagent with a bias greater than the maximum allowable bias.
A case study was discussed where a new lot of reagent showed a consistent bias in performance versus reference material in a head to head setting. Reference to the historical data indicated that the reference material showed signs of degradation due to long-term storage. Therefore, to plan for bridging of the next batches of reference reagents, accelerated stability assessments were conducted and a reference material stability monitoring plan was set up. This allowed the definition of a reference material expiry date based on data as well as a plan for bridging new lots of reference material. The lesson learned was that original reference material should be robust when used as a comparison.
In a second case study discussed, a new batch of pre-coated plates had to be evaluated in an assay used to support an on-going clinical trial in order to accommodate the expiration date provided by the vendor. Given that bridging of new plate lots consistently did not meet a priori criteria, the use of new plate lots would necessitate potential re-testing of clinical samples and partial re-validation. To mitigate the impact on on-going clinical sample analysis and allow time for assay re-validation, a long term stability experiment was conducted from historical plate lots. Confirming the historical accelerated stability data generated from the same lots, the data generated indicated that the shelf life of the plates could be extended beyond what was provided by the vendor. Therefore, when bridging between lots is not possible, consideration should be given to extending reagent stability. However, this must be anticipated by keeping reference material and generating reference data in due time.
Vaccine Bioanalytical Assays & Immune Monitoring
Vaccine assays must be well characterized, statistically supported and well controlled to measure safety and efficacy endpoints in vaccine trials. While safety, efficacy and good science are the bedrock for vaccine development, it is important to challenge the status quo to bring these life-saving medicines to improve human health. Hence, a bioanalytical scientist should think creatively to accelerate testing of life saving vaccines while maintaining safety and efficacy. This could translate to innovation in terms of platform selection and approaches to develop, optimize and validate these assays. Previous discussions suggested the consideration of new technology platforms to improve assay robustness and throughput given that vaccine assays need to be maintained over a long duration. For example, design of experiments (DoE) approaches were discussed to evaluate critical reagent pairings and optimization of assay conditions [25]. To reduce sample volumes needed, particularly in the paediatric population, technologies that enable analyte multiplexing should be considered even if switching from single analyte assays to multiplex assays has certain challenges (specificity and cross-reactivity of the reagents used must be evaluated) [28]. The 2019 White Paper in Bioanalysis also recommends that bridging with a priori acceptance criteria may be needed before switching to a multiplex format and a change in platform is not recommended for Phase III support as it may not be possible to demonstrate equivalency [25]. In addition, it may be necessary to reassess antigen concentrations and establish new reference standards [115].
Gold standard methods for vaccine antibody measurement tend to be low throughput, labor intensive, not amenable to automation and highly variable, adding to the timelines of bringing effective therapies to the populations that need it. ELISAs are often used as gold standard methods to detect the IgG-specific immune responses but are limited to the measurement of a single analyte and, in the case of multivalent vaccines, this means the use of multiple assays and large volumes of patient serum, a challenge for pediatric studies. This may often necessitate the use of multiplexing approaches or the use of novel technologies, which lends itself to multiplexing (increasing throughput and reducing sample volumes), is amenable to automation (increasing throughput) and has Laboratory Information Management System (LIMS) compatible software.
In addition to total IgG assays, it can also be important to have functional assays (i.e., NAb assays) to serve as a surrogate of protection. These often tend to be labor intensive assays with low throughput. Vaccine developers have been successful in introducing new multiplex methodologies such as plaque reduction neutralization tests (PRNT) [116,117], cassette-based infectivity assays and Virus Reduction Neutralization Tests (VRNT) [118,119].
To enable successful use of these novel technologies, regulatory agencies may offer consultations and scientific advice and sponsors are encouraged to take advantage of the service early in development. It was suggested that submitting the same questions to multiple regulatory agencies may encourage collaboration to minimize conflicting advice.
Health authorities prefer scientific rigor to provide assurance that the assay is fit for the intended purpose versus using a “gold standard” assay when it may not be the best scientific decision. Sponsors are also encouraged not to interpret guidance literally but to use it as a guide. All scientifically valid data should be included in the submissions.
Vaccine assays must be robust and rugged enough to endure several decades of clinical testing. Typically, vaccine assays have several phases that have been previously discussed in the 2019 White Paper in Bioanalysis [25]. Consensus was reached that if assays are used for safety evaluations and patient selection, they must be validated regardless of the development phase; assays used in support of pivotal studies must be validated prior to use to prevent basing decision on non-robust data.
Immune responses to vaccine administration can be highly complex requiring characterization of the innate immunity, B and T cell activation as well as antibody titers and functionalities. The complex immune response can be broken down into the early innate response and adaptive immunity. Early innate response is characterized by the measurement of chemokines and cytokines as well as cell mediated immunity. Adaptive immunity is characterized by measures of functional antibody responses, memory B cells and T cell responses, and B and T cell receptor (BCR and TCR) sequencing. The latter has enabled the cloning of receptors and the functional validation of a predicted specificity [120]. Methodologies such as high parametric flow cytometry, multiplexed assays, systems serology and transcriptomics are applied through preclinical and early clinical phases to provide an in-depth understanding of the vaccine mechanism of action (MOA).
In a study of the inflammatory parameters associated with systemic reactogenicity following vaccination with hepatitis B vaccines, the system reactogenicity was associated with innate immune response elements. These in turn were found to associate with the adaptive immune response, emphasizing the key role of the innate immune system in driving response to the vaccine [121].
In another study, to map the functional humoral correlates of protection against malaria following RTSS/AS01 vaccination, there was limited power to predict protection after vaccination when using individual antibody features. Specific multivariate functional signatures were associated with protection rather than antibody concentrations. A converging set of humoral correlates predicted protection irrespective of the vaccine arm. The arm specific correlates were found to predict protection across independent studies. Hence, using approaches such as systems serology profiling and controlled human challenge models may help define overarching correlates to define the protective efficacy of prospective immunogens and perhaps obviate the need for the more expensive Phase IIb/III vaccine studies [122].
The third case study discussed, described the evaluation of peripheral blood mRNA profiles and kinetics from tuberculosis vaccine candidate M72/AS01 using Affymetrix microarray and gene set enrichment analysis of RNA expression based on blood transcription modules (BTM). It was found that early response to the vaccine was characterized by the activation of several pathways linked to innate immune activation. This response was transient and succeeded by the activation of pathways related to the adaptive response. Evaluation of RNA expression of a set of genes related to the mode of action could be used to identify different responses to the vaccine using whole blood. Hence, systems biology has the potential to identify gene signatures associated with vaccine immunogenicity and protective efficacy [123].
A final case study discussed the use of high parametric flow cytometry and hemagglutination inhibition to study blood T follicular helper (Tfh) lineage cells which found that the early rise of blood Tfh cell subsets and baseline immunity were predictors of persistent late functional antibody responses to vaccinations in people [124].
Understanding the complexity of the immune response requires a suite of assays (transcriptomics, metabolomics, genomics, epigenomics, cell surface marker expression, antibody, cytokine concentration analyse) as well as the ability to evaluate multivariate data sets using a systems biology approach to refine vaccine development [125]. Systems biology tools can be harnessed to identify predictors of vaccine efficacy and discover new insights about protective immunity. This provides important knowledge that guides and facilitates development and eventually licensure of novel vaccines. It is also important to note that, as long as they remain exploratory, even if used on late phase clinical trial samples, these complex methods do not require to be fully validated. It is only if the outcome of one of these explorations is successful in identifying a potential biomarker (e.g., of protection) and these data are intended to be used to support the vaccine file, that the corresponding assay must be validated. In such a case, the validation of the assay can be performed a posteriori, provided it did not change between the generation of data and its validation.
RECOMMENDATIONS
Below is a summary of the recommendations made during the 15th WRIB.
TAb/NAb & Anti-Viral Vector Antibody Companion Diagnostic Assays
- Sensitivity of the NAb assay is dependent on the format, choice of reporter (green fluorescent protein vs luciferase), permissive cell lines, and starting dilutions.
- The impact of NAb on transduction is dependent on the AAV serotype, tissue of interest, target and ROA.
- Low titer NAb may not necessarily reflect clinical relevance and hence there may not be a need to constantly refine the NAb assay to improve sensitivity.
- TAb assays are most often used for ease of use, sensitivity, high throughput and association of TAbs with immunotoxicities. The need for a NAb assay against the viral vector and transgene protein is evaluated on a case by case basis. When determining whether to use a TAb assay or NAb assay to detect pre-existing immunogenicity for patient enrolment, it was recommended to consider the pros and cons of both formats as shown in Table 1.
- Prophylactic and in-treatment regimens with immune suppressive/modulatory treatments would not mitigate the pre-existing antibody response; however, they may diminish treatment boosted or emergent immunogenicity and broader impact on immune functions.
- When a sponsor believes that regulated clinically validated assay poses a non-significant risk (i.e., an incorrect test result does not pose a potential for serious risk to subjects in a trial) submitting a justification for this position will aid in the review of IND materials considering a totality of evidence approach.
- The use of one assay to support the development of multiple serotypes for multiple sponsors may be possible for the support of a clinical trial but not for a CDx.
- A CDx for one AAV GTx may be used as a LDT for the development of another AAV GTx of the same serotype if it can be justified.
- A given assay could be a CDx and not a CDx at the same time; the CDx is only required for patient selection and not for post-treatment sampling since the treatment-induced titer is not used for patient treatment decisions.
- It is important to engage with the FDA as soon as possible to determine whether or not a CDx is required for inclusion/exclusion of patients based on pre-existing antibodies.
Viral Vector Shedding Assays
- Even with ROAs where the probability of virus shedding is low (e.g., subretinal administration), viral shedding should be assessed with the focus on in vivo systemically administered therapies.
- Sponsors are encouraged to engage with the regulators and utilize data collected over the course of product development as justification for not conducting shedding studies.
- There is currently no regulatory guidance on the selection of analytical methodologies and how to conduct method characterization, validation and clinical sample testing for shedding assays; industry best practices provide guidance on the considerations and parameters that are important to characterize and validate qPCR/ddPCR methods [21,25,28,48].
- The current industry practice is to use qPCR/ddPCR methods for shedding assays and/or cell-based infectivity assays, if assay performance supports its use.
- The most commonly used matrices for shedding assessment are urine, nasal swabs, saliva, tears (for local eye delivery), and semen (adult patients only).
- When developing qPCR/ddPCR methods, the type of consent needed to collect samples, sample collection aspects due to degradation by endogenous enzymes and analytical platform sensitivity should be considered.
- Missing time points are a reality for these studies and the key challenge is in making these shedding studies as patient-friendly as possible while collecting the maximum amount data. Regulators are willing to accept these types of deviations to the study plan based on limited impact to available data.
- An assessment of the shedding of the empty AAV capsid is not recommended
Viral Vector Gene Therapy Immunogenicity & Pre-Existing Immunity
- Perform and update an immunogenicity risk assessment for GTx.
- It is widely accepted that current immunogenicity guidance documents [52,53] are broadly applicable for GTx, although the FDA guidance indicates that GTx are out of scope.
- Best practices include banking of samples, conduct of the immunogenicity assays at a single laboratory and evaluating the impact of immunogenicity on efficacy and adverse events.
- A cautionary approach which may require excluding patients positive for or with high titers of pre-existing antibodies is recommended for FIH studies.
- Consideration should also be given to immunomodulation approaches using corticosteroids to mitigate the side effects of the administration of a GTx and IgG depletion approaches to reduce pre-existing anti-AAV Ab titers to increase the potential benefit of these therapies.
- Industry practice has been the use of humoral pre-existing antibody responses for inclusion/exclusion criteria in trials (dependent upon the ROA).
- Prevalence of pre-existing antibodies to AAV varies by serotype, age, and geography, therefore, it is important to understand the impact of pre-existing antibodies on clinical outcomes.
- Immune modulation as well as adjustments to dosing and IgG removal were proposed as options to re-administer GTx to seropositive subjects.
- IgG removal options include the use of transient enzymatic degradation or the mechanical removal of pre-existing antibodies by plasmapheresis or immunoadsorption.
- Understanding the mechanisms behind immune response is key to designing an efficient immunogenicity mitigation strategy and monitoring for AAV based GTx.
- The timeframe for immune response monitoring (humoral or cellular) to assess safety and efficacy includes:○pre-treatment (i.e., screening stage) – for pre-existing anti-AAV antibodies.○immediately prior to administration – to determine whether seroconversion has occurred between screening and dose administration.○post-treatment – cellular and humoral responses to the vector and transgene product.
- There are not enough data to conclusively link pre-existing antibody to efficacy in part because of the lack of standardization amongst assays.
- Seek regulatory advice early to determine the appropriate assay to assess pre-existing antibodies with the understanding that the impact of pre-existing antibody is more nuanced than magnitude (titer) of response.
- Since a qualitatively similar immunogenicity response may have different in vivo responses, consensus was reached that a universal threshold of pre-existing antibody set as an inclusion/exclusion criteria is not relevant.
CRISPR/Cas9 Immunogenicity & Bioanalytical Challenges
- The promise of the CRISPR-Cas9 gene editing therapy lies in the ability to exploit the cell’s DNA repair machinery by introducing changes to one or more genes.
- Currently, there is insufficient data to draft a guidance, but it was recommended to generate data to support patient safety by the use of in vitro assessments as well as the use of animal models while acknowledging that these may not always translate in the clinic. These studies could include the use of T cell receptors on CD-4 cells and studying protein engagement using MHC Class I molecule presenting cells.
- Pre-screening for Cas9 protein antibodies may not be feasible for orphan or rare disease populations.
CAR-T Immunogenicity & Cellular Kinetics
General Considerations
- As autologous T cells present a low to medium immunogenicity risk, an anti-CAR antibody assay is typically implemented, often starting with the FIH study. Anti-CAR antibodies may be detected using classic LBAs.
- The use of engineered surrogate CAR-T cells or target cells carrying the reporter gene may overcome some reagent supply challenges.
- For autologous bi-paratonic CAR therapies:○it is important to assess immune responses to the CAR ECD as well as the individual domains.○a multiplex platform allows for the simultaneous assessment of NAb to the full-length CAR and two separate binding domains, respectively.○matrix interference must be mitigated.○parameters like drug tolerance, sensitivity, and precision should be evaluated.○approaches to improve drug tolerance may also help with reducing matrix interference.
- There was clear consensus that assay controls are to be implemented periodically to track that the assay is performing consistently [25,28].
- To overcome the challenge of limited availability of patient-specific T cells, CAR transduced cell lines that offer high transduction rate and consistent CAR expression can be used in combination with anti-human IgG/IgM or anti-human IgG Fc. Adequate justification for cell line selection should be provided.
- Advantages to using flow cytometry:○CAR proteins do not have to be labelled minimizing the risk of masked epitopes.○not limited due to the presence of insoluble/hydrophobic domains.○presents CAR processed through the T cell expression system (e.g., glycosylation).○allows for CAR presentation in the native environment.○native conformation on cell surface along with potential interaction partners.
- The use of wild type T cells may be needed as a control to identify (and possibly exclude) non-CAR specific responses.
- Cell populations may “shift” over time in the FSC/SSC (forward scatter/side scatter) plot, necessitating the setting of appropriate gates and performing daily QC performance checks of the cytometer.
- Minimizing the inclusion of mouse components when designing CAR-T therapies may be important.
- Consider patient specific cut points – using the patient pre-treatment samples to calculate the cut point factor. Statistical approaches may be considered. Regulators are looking for justification of the approach used.
Cellular Immunogenicity Assays
- ELISpot and FluoroSpot are platforms available for assessment of cellular immunogenicity.
- The cellular immune response against CAR can be assessed using a patient’s PBMCs.
- ICS can also be used as a functional measure of antigen-specific cellular immune response to CAR.
- Development of ICS assays:○use of sodium azide and low temperatures can prevent internalization of surface antigens.○Fc receptors can be blocked using specific reagents such as Fc block.○complements can be reduced by heat inactivation.○the antibody selected must not bind the scFv portion of the CAR directly.○the use of Fab(2) detection antibodies may help mitigate the potential to bind Fc receptors.
Bioanalysis of Cellular Kinetics
- Direct and indirect approaches can be used to measure CK:○Flow cytometry assays quantify CAR-T cells directly and can also provide information on the distribution of CAR expression.○qPCR/ddPCR assays that quantify the copy numbers of transgene can serve as an indirect measure.
- The choice of methods should reflect the consideration of the research question at hand as well as the pros/cons of each method.
qPCR/ddPCR Assay Performance
qPCR
- One of the parameters that introduces variability in the RT-qPCR methods is the efficiency of the RT step.
- RT efficiency can be impacted by the mRNA encoded by different genes, mRNA integrity and purity, the enzyme efficiency and primer selection strategy.
- To determine RNA integrity and purity, the RIN can be used as a measure.
- Using RNA standards enables absolute quantitation by accounting for RT efficiency and variability and will allow for comparison of transcript levels across studies.
- Using dsDNA as standard material does not account for the RT efficiency and variability and is not considered absolute quantitation.
- For QC selection/preparation, spiked RNA can be used, however this approach does not reflect the RNA recovery from the homogenization/cell lysis process and the degree of degradation of target RNA during this process cannot be accurately measured.
- Absolute quantification method:○use the standards and controls prepared in matrix RNA from relevant species and tissues.○mRNA should be reported as copy/μg of total RNA and not normalized to mass of tissue.○assay parameters selected for evaluation are based on those routinely used for LBAs:▪range of quantitation.▪sensitivity (copy/reaction).▪selectivity (at least 8/10 individual matrix RNA spiked with QC should recover within 80–120%).▪precision and accuracy.▪stability.
- Relative quantitation of gene expression using the ΔΔCt, method:○Use RG genes to control for sample input; they should be constantly expressed with a high amplification efficiency, and be unaffected by treatment.○There should be equal amplification efficiency (≤5%) between the target gene and RG gene(s).○QCs are needed to monitor run to run performance of these methods.○Prepare QC by diluting the total RNA isolated from relevant naive species/tissues in carrier RNA and determine precision in multiple runs (e.g., n = 5).○%CV should be ≤2%Ct.○when normalized to RG genes should recover within ±25% of the undiluted QC.
- Characterization of viral vector shedding method:○A FFP validation approach using standards, QC and NTC is recommended.○Calibrator material should be equivalent or very similar to test samples.○If the assay is intended to be used over an extended period of time, it is preferred to use cGMP material.○If research grade material is used, bridging to the lot of clinical material should be considered as soon as possible.
- qPCR method development:○when selecting primers and probes:▪avoid primer/primer dimers.▪consider physical properties.▪consider amplicon length.▪efficiency: 90–110%, 85–90% may be acceptable for early studies.▪sensitivity and specificity.▪ideally select such that the same pair can be used across pre-clinical and clinical studies.○identify primary and secondary tissues/biofluids of interest.○start with as high a recovery as possible: 30–80% with >50% for primary tissue/biofluids.○refer to Table 3 for updates to prior recommendations for all validation parameters.
qPCR in vaccine development
- Vaccine assays:○Parameters used for vaccine serology assays [28] are still relevant (precision & accuracy, robustness & ruggedness)○LOD:▪evaluated using a series of dilutions of the antigen in negative samples▪%positive results (# positives/total*100) is plotted and fit with an appropriate regression model▪is the copy number at which the model predicts the assay can detect 95% of the time○For qualitative PCR assays, it is adequate to assess the linearity and dynamic range.○For qPCR assays, it is important to establish the LLOQ using WHO reference standards when available.○False positive probability is an essential part of the validation:▪use a large number (>1000) of confirmed negative samples interspersed with positive samples.▪use of engineering controls from purification to PCR analysis should be considered to minimize the risk of contamination, similar to those used for RNA analysis▪positive and negative samples are purified and processed in an identical manner▪a binomial distribution is used to calculate the upper bound on a 95% confidence interval▪if the numbers of positive and negative samples are small, a point estimate can be considered
ddPCR
- To characterize CAR-T cell distribution, expansion, contraction, and persistence profiles and facilitate data comparison across different studies, it is recommended to develop a universal duplex ddPCR assay protocol.○utilize a single copy gene as the reference gene to normalize the gDNA input.○the primer/probe for the reference gene should be designed to target conserved DNA sequences in the genome (e.g., between mouse and human) to facilitate data comparison.
- QC samples can be prepared by spiking plasmid containing a species-specific CAR target sequence in mouse gDNA (pre-clinical studies) or human gDNA (clinical studies).
- For each 96-well plate run, at least two sets of QC samples should be included and tested in triplicate.
- The following parameters of assay reliability are recommended for assay validation: precision and accuracy, sensitivity, selectivity, LLOQ, and short-term stability (freeze/thaw, refrigerated stability).
- Provide a rationale for the approach to validation.
- Refer to Table 5 for acceptance criteria for ddPCR assay validation and Table 6 for plate acceptance criteria.
- Use the mean of at least 3 repeat measurements, where each measurement is performed in triplicate, to determine the nominal value of the prepared QCs.
Analyte Stability in Vaccine Serology Assays
- Demonstration of long-term stability of immune response to vaccine-induced antibodies may not be necessary given that there is evidence documenting stability of antibodies (including IgG and IgM isotypes) in frozen serum at less than -60°C for up to 4–5 years.
- Proficiency panel data can be used as an alternate approach to assess long-term stability.
- When samples require additional processing, stability assessments should cover the entirety of the sample handling/sample processing steps.
- QC trending may be performed using tools such as an exponentially weighted moving average.
- Method parameters that may not necessarily have acceptance criteria (e.g., cell viability, slope, functional output) can also be used to monitor trends in method performance and drift.
Vaccine Critical Reagent Management & Bridging
- Critical reagent bridging is recommended prior to using a new critical reagent lot.
- When designing appropriate bridging experiments and defining acceptance criteria, it was recommended to understand the intended use of the assay and assay performance.
- Critical reagents should first be evaluated for their biophysical and chemical properties to develop a set of replacement criteria which includes performance characteristics, based on their use:○this can be followed by comparison of assay performance parameters using current and new reagent lots in a head-to-head comparison.
- Ideally, a panel of incurred samples is compared head to head with the candidate and the qualified reagent in multiple independent runs. If not possible, the performance of QC and mock samples should be assessed.
- Assay trending of reagent dependent parameters may be used to evaluate critical reagent bridging.
- If a qualified lot of reagent is not available for comparison purposes (e.g., expired), a comparison to historical data generated with the qualified lot may be considered.
- For each assay and reagent type, define detailed bridging design and criteria:○optimal sample panel size range (including a minimum number with valid final results).○distribution of samples (including a range of responses).○appropriate statistical parameters to be analyzed and acceptance criteria.○potential impact of accepting a reagent with a bias greater than the maximum allowable bias.
- Original reference material should be robust to the storage conditions and time stored prior to use when used as a comparison in bridging experiments.
- When bridging between lots is not possible, consideration should be given to extending reagent stability but this requires anticipation.
Vaccine Bioanalytical Assays & Immune Monitoring
- Multiplexing approaches or novel technologies should be considered for vaccine assays testing serotype specific IgG responses or functional activity to increase throughout and decrease sample volumes.
- When using novel technologies, regulatory agencies may offer consultations and scientific advice and sponsors are encouraged to take advantage of the service early in development.
- If assays are used for safety evaluations and patient selection, they must be validated regardless of the development phase.
- Assays used in pivotal studies must be validated to prevent basing decisions on non-robust data.
- Submit the same questions to multiple regulatory agencies to encourage collaboration to minimize conflicting advice.
- Health authorities prefer scientific rigor to provide assurance that the assay is fit for the intended purpose versus using a “gold standard” assay when it may not be the best scientific decision.
- Sponsors are encouraged not to interpret guidance literally but to use it as a guide.
- All scientifically valid data should be included in the submissions.
- In the early stages of vaccine development, when understanding the mechanism of action is important, one should consider a systems biology approach and integrate data from a variety of tools (RNA expression, humoral and cellular immune responses) to understand the complex immune response.
SECTION 2 – Immunogenicity of Biotherapeutics
Susan Kirshner23, Daniela Verthelyi23, Haoheng Yan23, Kimberly Maxfield23, Joao Pedras-Vasconcelos23, Mohsen Rajabi Abhari23, Swati Gupta29, Yuling Wu30, Manoj Rajadhyaksha31, Matthew Andisik17, Daniel Baltrukonis32, Elana Cherry24, Isabelle Cludts33, George Gunn34, Anders Holm Millner35, Gregor Jordan36, Sumit Kar22, Robert Kubiak30, Gregor P Lotz36, Rachel Palmer37, Kun Peng38, Johann Poetzl39, Susan Richards37, Natasha Savoie22, Roland F Staack36, Kay Stubenrauch36, Meenu Wadhwa33, Günter Waxenecker40, Tong-Yuan Yang4 & Lucia Zhang24
Authors are presented in alphabetical order of their last name, with the exception of the first 9 authors who were session chairs, working dinner facilitators or major contributors.
The affiliations can be found at the beginning of the article.
HOT TOPICS & CONSOLIDATED QUESTIONS COLLECTED FROM THE GLOBAL BIOANALYTICAL COMMUNITY
The topics detailed below were considered as the most relevant “hot topics” based on feedback collected from the 14th WRIB attendees. They were reviewed and consolidated by globally recognized opinion leaders before being submitted for discussion during the 15th WRIB. The background on each issue, discussions, consensus and conclusions are in the next section and a summary of the key recommendations is provided in the final section of this manuscript.
NAb Assays – Drug & Target Interference
What are the current performance expectations for NAb methods? Are guidances used for ADA applicable and appropriate for NAb? What are the key parameters of interest (sensitivity, drug tolerance, long term robustness)? What is the minimal amount of critical reagent characterization? During method development should one focus on sensitivity or drug tolerance? Is it really necessary to have a NAb assay and bioanalysis for low-immunogenicity risk molecules? Could a sponsor have a conversation with regulators for an integrated approach or accept the risk for a postmarketing commitment (PMC)?
ADA Cut Points – Appropriateness & False Positive Rates
What are the approaches to correlate statistically determined CP versus specificity? Given that commercial samples are generally different from study population should other approaches be considered?
Circulating Immune Complexes – ADA/Drug Complexes
What are the new developments in circulating immune complexes (CIC)? How do CIC impact clearance in preclinical species?
ADA Assay Comparability
What is the regulatory perspective on platform agnostic methods and comparison of immunogenicity incidences? What are the needs and considerations for ADA assay cross-validation? What are the differences in approach for low risk versus high risk molecules?
Integrated Summary of Immunogenicity Harmonization
What are the core elements of the Integrated Summary of Immunogenicity (ISI) that should be harmonized? What is the content in the report? What is the expectation for Section 2.7.2.4 if ISI is provided in Module 5?
China NMPA Immunogenicity Guidance
Can the China NMPA Immunogenicity Guidance be harmonized with the FDA and EMA guidance/guideline [52,53]? Is it necessary to validate non-clinical ADA assays? What are the FDA and EMA regulatory insights on ADA assays for GLP tox studies? Would these studies need to be validated if molecules have demonstrated cross-reactivity with NHP and rodents?
Multi-Domain Biotherapeutics & Bispecific Antibody Immunogenicity
Do we need a confirmatory assay to understand domain specificity? Should any additional characterization be done post confirmatory assay for the entire molecule? Are domain specific PCs needed? What is the general opinion on the use of surrogate ADA isotype (purified ADA positive samples, chemically conjugated ADA-idiotype antibody with isotype or recombinant ADA)? Are all approaches equally suitable? What is recommended?
Biosimilar ADA Assay Validation & Harmonization
Can consensus be reached for criteria to assess suitability of the one assay approach using immunodepletion curve analysis? What strategy should be used to execute method comparability experiments? Are there additional factors needed to support the one assay approach?
DISCUSSIONS, CONSENSUS & CONCLUSIONS
NAb Assays – Drug & Target Interference
NAb assay formats are typically evaluated in alignment with the immunogenicity risk assessment, with cell-based bioassay formats generally preferred by most regulatory agencies. When appropriate, using the competitive ligand binding assay format is acceptable [110]. Cell-based bioassays provide experimental systems that reflect all or key portions of the pharmacologic activity of the therapeutic. Competitive LBAs demonstrate inhibition of drug binding to its target. However, both cell-based and competitive LBAs are susceptible to drug and/or target interference [21,24,25,28,52,53,110].
Previous White Paper recommendations have discussed how and when to apply the different NAb formats to characterize ADA neutralizing capacity, primarily guided by therapeutic immunogenicity risk, MOA and modality. In general cell-based bioassay formats are thought to better represent the pharmacologic activity of the drug; however, competitive ligand assay formats may be appropriate for antagonistic drugs or when a cell-based assay is not feasible. The 2015 White Paper in Bioanalysis acknowledged potential for both false positives and false negatives in NAb assays due to drug or target interference. Acidification, blocking/binding agents and pre-treatment to remove the interfering molecules were some of the solutions discussed along with potential caveats associated with these approaches [12]. The 2019 White Paper in Bioanalysis defined questions that remained regarding NAb assay performance and expectations. Discussions of matrix interference (e.g., drug) were primarily focused on ADA methods [25]. Regulatory guidance recognizes the complexity of NAb assays [52,53,110]. Robust assay methods are essential to enable smooth routine sample analysis. Fewer assay steps would reduce the complexity and overall duration of the assay. This update of these evolving recommendations is focused on discussing current regulatory and industry guidance and expectations.
In general, cell-based NAb methods are unlikely to match the performance of ultra-sensitive bridging ADA assays, but there may be opportunities to improve their performance to better enable effective characterization of neutralizing capacity. There are advantages to characterizing interference of cellular events versus solely antibody binding. Strategies to maximize NAb performance include assay platform optimization and sample pre-treatments adapted from ADA assays. Acid dissociation, ACE, solid phase extraction with acid dissociation (SPEAD), biotin-drug extraction and acid dissociation (BEAD), and precipitation and acid dissociation (PandA) have all been used [80,126–128].
These assay formats might also deliver false positive or negative assay results in the presence of soluble target which may interfere with the assay readout especially at high target concentrations. The observed interference could be related to either homodimerization of the soluble target or aggregation during sample processing. In the case of a premedication with a biologic targeting the same soluble target (e.g., with an IgG-like molecule binding a different epitope than the substance tested), a false positive assay result may also be observed. As more biologics become available, targeted premedication may be observed more frequently and has to be addressed during bioanalytical assay development in terms of matrix interference. Essentially, two procedures come into question; one to significantly reduce the soluble target concentration and the other to mask all but the epitope of interest.
Testing strategies were discussed to address the challenge of high concentrations of soluble target. One example highlighted the challenge of selecting screening assay conditions that preserved the positive assay signal to allow setting a scientifically sound cut point for the confirmation step. The other example focused on a soluble target depletion step that could be of value for other projects as no specific reagents were required.
The fundamental parameters for ADA and NAb assays validation include (1) cut point; (2) sensitivity and drug tolerance; (3) specificity and selectivity; (4) precision; (5) reproducibility (when relevant); (6) robustness; and (7) in-use stability of critical reagents [52]. The USP guidance [110] was also suggested as another possible resource for NAb assays validation.
Cell-based assay sensitivity is generally much lower (μg/ml levels) compared to competitive LBAs (ng/ml levels) that can approach the sensitivity expected per guidance (≤100 ng/ml). Regardless of the method used to improve assay sensitivity and/or drug tolerance, it is key to demonstrate adequate assay sensitivity in samples with levels of drug anticipated at sampling time points. Regulators understand the compromises needed to balance sensitivity and drug tolerance of a method and the assay parameter assessment depends on the surrogate positive control used in assay development. Sponsors should summarize the efforts conducted and focus on a FFP philosophy when developing and validating assays. An assay should, at a minimum, be validated at the drug trough level. In addition, an integrated approach can be considered with the totality of the evidence in cases where target assay drug tolerance is not reached. It was also noted that regulators may ask for data for “wash out” samples for interpreting patient response to the drug and duration of efficacy.
The goal of an immunogenicity assessment is to evaluate the clinical impact of ADA on safety/efficacy, including impact on PK/PD, with an assay adequately developed to aid this goal. To do so, regulators look at the totality of evidence. For some modalities where NAb data are limited, regulators may require a PMC. For certain molecules and indications, there may be results from other assays that indicate neutralizing activities (e.g., PD/TE, free PK, biomarkers). There is a variety of possible situations where the totality of evidence may be considered to detect product neutralization in a clinical study such as the presence of an endogenous counterpart or assays that are validated with surrogate controls not reflective of sample response.
Consideration of strategies using the integrated data approach and totality of evidence with regards to developing dedicated NAb assays were discussed. The evaluation of the neutralizing effects of ADA is crucial when characterizing immunogenicity and potential effect on efficacy and safety of a therapeutic drug. The need and timing for an in vitro NAb assay in the clinical program should be based on the immunogenicity risk of the drug. Using in vitro assays to measure the ability of ADA to block binding between the drug and its target is the current recommended approach for pivotal trials [52,53]. However, increased focus is being directed towards an integrated data approach where magnitude of ADA response is correlated to efficacy and safety instead of using a standalone in vitro NAb assay. There was consensus that immunogenicity assessments integrated into PK/PD can significantly benefit regulatory assessments.
To improve the application of the integrated data approach, there is a need for more aligned guidance. Recommendations on the immunogenicity integrated data approach started in 2017 and they were further expanded and updated based on new case studies and acquired knowledge in 2018, 2019 and 2020 [18,21,25,28]. The challenge with using the traditional in vitro NAb assay is the subsequent translation of the complex, positive/negative, transient/persistent classification patterns into clinical relevance. NAb assays are not always conclusive and have limitations to provide information about the clinically relevant levels of NAbs. Conversely, the integrated data approach evaluates impact on exposure, efficacy and safety by correlating titers and persistence of ADA to PK, PD, biomarkers and safety events and can also include relevant in vitro NAb assay data. Clinical relevance of the preponderance of ADA/NAb data is important because NAbs may have a large clinical impact (as no large immune complexes need to be formed) even at low concentrations (as the antibody:drug ratio for inactivation is close to 1). Immune responses with drug clearing potential may be polyclonal with high titer or target a repetitive antigen and be in excess to drug; however, there are indications that the morphology of the formed complexes, even with monoclonal antibodies (mAbs), might drive the clearance [129,130]. Recent studies indicate that increased clearance of immune complexes (ICs) is not an ADA property, but rather every ADA (monoclonal and polyclonal) can have clearing potential when the formed ICs have a critical size (dependent on ratio and concentration of drug and ADA) in combination with the structure of the formed ICs [130].
A case study using the integrated data approach rather than an in vitro NAb assay for the assessment of clinical relevant ADA was discussed for an investigational drug with low immunogenicity risk and a PD biomarker that reflected the presence of the active drug target (ligand) [18,21]. The mAb targeted a soluble ligand, with no risk of ADCC. The drug was used in addition to standard of care with no adverse events related to ADA in Phase II. There was low frequency of ADA formation in Phase II, with one case (<1%) with an effect on PK and PD. At the end of Phase II meetings with the US, EU, Japan, and Canada health agencies, an integrated data approach to assess NAb in Phase III was proposed. The ligand binding PK assay, using a bridging format, likely demonstrated ADA interference. The ADA assay (bridging format) did not demonstrate drug interference at the relevant drug concentration. It was hypothesized that the long drug half-life at μg/ml levels likely compromised in vitro NAb during and post treatment. ADA, PK and PD samples were analyzed at the same time points and even in the absence of reliable NAb data, the results clearly indicated that high ADA titers led to reduced PK, which in turn led to alteration of PD. Therefore, it is important to consider if the PK assay can measure functional drug in the presence of ADA or if it is inhibited by ADA. In selected cases, where there is a highly sensitive PD marker and/or an appropriately designed PK assay that generates data informing clinical activity, it may be possible to use these data in lieu of a NAb assay, particularly for low-immunogenicity risk molecules. This determination should be done in consultation with the regulators.
Regulators provided feedback to this case study, stating that one example to evaluate the integrated data approach may not be sufficient to build a generalized case for this and other molecules. The PD marker was not the only downstream effect caused by removal of the ligand, hence the drug effect could not exclusively be related to the suggested PD marker; other factors may have influenced the PD marker. While true, those factors are typically transitory, and PK and ADA are still available to inform. If the ligand activity can be directly measured by the PD marker, other PD markers may not be required. It was also recommended to include free ligand (e.g., target engagement assay) and clinical effect in the integrated data approach. However, the free ligand was depleted in vivo within minutes and extensive accumulation of ligand on the drug was observed. Both factors complicate the measurement of free ligand levels. Additionally, dilutional linearity for target measurement was compromised due to a change in equilibrium during dilution between ligand and drug (dilution-induced complex dissociation) resulting in a bioanalytical challenge during assay development and platform selection. Finally, it was proposed to address the clinical consequence of ADA, which in rare event studies, can only be addressed at the population level.
There have been proposals for tiered approaches for drugs of varying immunogenicity risk where an integrated or traditional approach could be used. For example, the 2019 White Paper in Bioanalysis [25] recommended an approach to catagorize ADA using an integrated data approach: 1) non-neutralising: ADA with no impact on exposure levels or clinical efficacy; 2) neutralizing activity that is not clinically meaningful: ADA causing reduced exposure, but no impact on clinical efficacy; 3) presence of clinically meaningful neutralizing activity: ADA leading to reduced exposure and loss of efficacy. Regulators confirmed that they are currently requesting the inclusion of in vitro NAb data, even for lower immunogenicity risk drugs, to enable them to, in general, gather information on the future usefulness of in vitro NAb assessments. However, for very low-risk molecules, where the immune response does not have an impact on PK/PD/AE, a NAb assay may not be needed. Sponsors should discuss with regulators program-specific approaches which differ from the recommended tiered immunogenicity analysis that include the use of a dedicated NAb assay. Regulators review each molecule on a case-by-case basis and immunogenicity data gaps will result in requests for additional data at the pre-approval or, sometimes, in post-approval stage as a PMC.
ADA Cut Points – Appropriateness & False Positive Rates
A key parameter in the validation of ADA assays is the determination of assay cut points, which are thresholds that define a sample result as negative or positive. Per regulatory expectations and industry practice, cut points are statistically determined using samples from treatment-naive subjects to yield false positive rates (FPR) of 5% and 1% for screening and confirmatory assays, respectively [28]. A common approach for determination of cut point appropriateness is through the verification of achieved false positivity during validation and sample analysis.
The original recommendations from Shankar et al. [106] were predicated on experience with monoclonal antibody therapeutics, interferon products, and replacement enzymes. The multi-tiered testing strategy and statistical cut point approach was initially recommended to assure true low positive responses can be detected in the assay. While many current cut point approaches are variations on Shankar et al. [106] and are well established, the calculation of cut points continues to be discussed. A conservative approach yielding false positives that can be further evaluated for specificity, rather than risking a false-negative result, has been the preferred approach so far.
One approach currently being discussed is using mathematical modeling for CP determination. Given the evolution of industry experience with cut points, regulators discussed the use of mathematical models for cut point assessment in certain populations (e.g., pediatric populations, rare diseases) but concluded that there often is not enough data to prove the validity of this method. Thus, while it may be possible going forward, additional data will be needed to establish the approach.
Current assay technologies, formats, and sample pretreatments beyond acid dissociation have led to significant improvements in both assay sensitivity and drug tolerance, resulting in highly sensitive assays [80,126,131,132]. As a result, low amounts of ADA are detected more frequently; however, the clinical relevance of low amounts of ADA continues to be discussed.
False positive rates in the validation and sample analysis phases are the simplest methods for verifying cut point appropriateness. Alternative approaches were discussed. According to industry, too much emphasis has been placed on achieving the nominal FPR. It is critical to understand the relationship between cut points and sensitivity/specificity in addition to false positivity, and it was proposed to consider cut points derived from sensitivity and specificity data. It was postulated that cut point appropriateness can be demonstrated by exceeding assay variability (minimum/precision cut point) [133], achieving acceptable assay sensitivity (e.g., ≤100 ng/ml), and using the assay specificity curve and visualizing/gating the data. Titers reported as signal-to-noise ratios (SNR) may help to provide additional insight into cut points.
Case studies were used to debate the relationship between false positives, sensitivity, and specificity, discuss new ways to visualize cut point data and appropriateness, and put these data in the context of clinical relevance [134]. These case studies suggested that statistically determined cut points based on false positivity have improved ADA detection. For example, one case showed that when determining a cut point while optimizing drug tolerance did not meet false positive expectations, it could still potentially achieve desired specificity and adequate sensitivity needs. Another case study also highlighted the possibility of an alternative transformation of screening data to reduce the FPR [135].
Other case studies describing simulations of immunogenicity testing for populations of various sizes and different proportions of “true” ADAs were discussed. These suggested that the observed ADA incidence can vary significantly depending on the sample size and applied false positive rate. These simulations showed that using a 1–5% FPR for non-immunogenic drugs is likely to result in assays with a low positive predictive value (PPV) (i.e., with a low (<50%) probability that a positive classification is truly positive). Assays with low PPV generate an overly large proportion of false positives which can dilute relationships between ADA and PK/PD, efficacy, and safety.
The possibility of erratic ADA incidence provides an argument for lowering the required FPR for biotherapeutics with low immunogenicity. It also underscores the weakness of reporting immunogenicity in terms of incidence since this number alone does not inform clinicians about the actual impact of ADA. ADA titer is typically more predictive of impact on PK/PD however, meaningful interpretation of titer (high versus low) is also challenging due to different reporting methods. It was suggested that sponsors could perform a tertile or quartile analysis of titer and outcome measures to determine interpretation of clinical impact. The challenge for industry is to move beyond reporting ADA as incidence and develop more quantitative methods for evaluating immunogenicity. It was discussed that, in certain cases, using SNR instead of titer values could also support the determination of the magnitude of ADA responses and may have greater precision than titer numbers for low magnitude ADAs detected by highly sensitive methods.
The approaches used to correlate statistically determined cut points versus specificity were discussed. Given that commercial samples are generally different from the study population, other approaches may need to be considered (e.g., study specific cut point). A sponsor can use mathematical models to determine experimental conditions but there is currently insufficient evidence to support this approach and sponsors need to provide data as a justification if using this approach. While it may be possible going forward, it may require a different paradigm for review.
Circulating Immune Complexes – ADA/Drug Complexes
There is major interest within the industry and from the regulators to better understand how the administration of therapeutic proteins might evoke the formation of ADA leading to the formation of ADA-drug ICs that may impact safety, efficacy, and PK. Clearance of ICs depends on the size and lattice structure [136–139].
The discussion expanded and deepened the 2020 initial recommendation on the generation of characterized ADA-drug ICs based on newly acquired experience in this field [28,140]. Case studies from in vivo PK studies performed in rats using pre-formed IC were discussed [130]. Different test compounds can be produced for IC PK studies to understand the influence of Fc-effector function, epitope specificity, and clonality [141]. For a detailed evaluation of how ICs with different compositions, size, and properties impacted drug PK, different ADA surrogates (mAb, polyclonal antibody, different epitopes) and therapeutic mAbs with different Fc-effector functions were used. In addition to the evaluation of the effect of IC formation on the total drug PK, an IC-size specific PK analysis was performed. It was shown that there were no significant differences between serum and plasma; there was no loss of ICs during serum preparation. Serum was deemed a suitable matrix for IC analysis and immunogenicity testing in general and thus all further analyses were performed in serum. Total drug PK ELISAs were performed for in vivo studies. These studies led to conclusions on the impact of defined IC sizes and structures and Fc-effector function on drug PK. The studies indicate that increased clearance of ICs is not an ADA property, but rather every ADA (monoclonal and polyclonal) can have clearing potential when the formed ICs have a critical size (dependent on ratio and concentration of drug and ADA) in combination with the structure of the formed ICs [130]. IC formation also results in enhanced proteolysis and renal excretion.
New advancements in CIC bioanalysis were discussed. The methodology is a significant technological advancement and new data have been generated to better understand ADA:drug IC impact on clearance. Pre-analytics–Size exclusion chromatography (SEC) coupled with ELISA has been used to determine the complex size distribution. The integrated data approach can also be valuable. However, the pathology and impact of CIC have not been evaluated in pre-clinical studies and species and needs further investigation and discussion.
ADA Assay Comparability
Life cycle management of immunogenicity assays for biologics often involves the development and validation of multiple immunogenicity assays at different stages of a clinical program. The pace of assay development is typically dictated by multi-disciplinary immunogenicity risk assessments in the early stages of the individual drug program. Some drug programs have complex development histories, with clinical study samples tested using different assays developed by different CROs. This can result in very challenging immunogenicity assay assessments by regulators to support the use of the totality of the immunogenicity data for proposed indication(s).
Currently, ADA assay comparability is not possible due to differences among the bioanalytical methods used. At the 13th WRIB, an approach to attempt the comparability of bioanalytical methods for immunogenicity assessment was proposed for the first time, including how to implement this approach in ADA assay development, how to report immunogenicity data to allow comparison across drugs, and how to use bioanalytical method comparability as the first potential step toward the evaluation of clinical immunogenicity comparability [25]. The discussion on novel approaches to ADA assay comparability continued amongst industry participants in 2020–2021 at multiple conferences but recommendations were never issued on this hot topic. To further continue the discussion, it was considered essential to include the expert opinion and input of regulators since no official position has been provided so far and no assay comparability data has been submitted to regulatory agencies at the time of this publication.
The below portions of the 2019 FDA immunogenicity guidance stating the limitations in comparing ADA incidence across therapeutic protein programs were discussed [52]:
FDA cautions that comparison of ADA incidence across products, even for products that share sequence or structural homology, can be misleading because detection of ADA formation is highly dependent on the sensitivity, specificity, and drug tolerance level of the assay [52]”.
“Therefore, comparing immunogenicity rates across therapeutic protein products with structural homology for the same indication is unsound, even though fully validated assays are employed [52].”
“When a direct comparison of immunogenicity across different therapeutic protein products that have homology– or across similar therapeutic proteins from different sources– is needed, the comparison data should be obtained by conducting a head to-head clinical study from which samples obtained are tested using an assay demonstrated to have equivalent sensitivity and specificity for antibodies against both therapeutic protein products [52].
Regulators do not recommend using platform agnostic methods to compare immunogenicity rates across different drug programs produced by different sponsors. A valid comparison of the technical performance of ADA assays would be an important first step that requires a joint effort of the entire community, from industry to regulators. To stimulate discussion, the use of the amount of cut point-ADA reagent complexes (CP-ARC) as a defined, PC-independent assay parameter that enables comparison of technical assay performance was proposed [142]. No recommendations were issued from the 13th WRIB and 14th WRIB on the CP-ARC concept, however new data and case studies were discussed.
The CP-ARC is a constant assay performance parameter that defines the assay sensitivity and drug tolerance [142]. CP-ARC is defined as the amount of ADA/analyte reagent complex required to give a signal at the assay cut point. The critical steps and challenges were discussed as well as the current gaps that need discussion and consensus within the scientific community. Real life challenges include the presence of residual drug requiring dissociation [80,126,131], dilution or reagents for new binding partners. Other challenges include interaction between ADA and drug/reagent. The interaction between ADA and drug and the interaction between ADA and labelled drug/reagents should be identical. Knowledge of in-solution binding kinetic information is an asset. Assay conditions and reagents should be appropriate to form defined ARCs. Interactions between ADA, PC and reagents need to reach equilibrium. Finally, correct concentrations of (active/binding competent) reagents and PC are required.
The potential benefits of this approach are improved comparison of ADA-assay performance and improved comparability of the clinical immunogenicity of different drugs. Another benefit is improved ADA reporting (i.e., no inconclusive results), since assay performance is fully understood. A clear statement of the detectable ADA concentrations is also an advantage. Joint efforts of the entire community, from assay development scientists to regulators, are needed to provide additional data, examples and experience to further support the CP-ARC concept. This includes PCs with different affinities, technologies, IgM-ADAs, and assay formats. Evaluation of the indicated existence of an assay-specific affinity cut-off for assay sensitivity (reagents and conditions) and a better understanding of the binding properties of ADAs are also needed. Regulators recommend that sponsors considering these alternative analytical approaches for their immunogenicity programs discuss them with regulatory agencies as early as possible.
Industry best practices for ADA method transfers may be required during the lifecycle of a program (e.g., to support analysis in China or transition from internal to external laboratories). Regulators recommend that when different sets of assays are used within the same immunogenicity program, data supporting labelling should be based on the analysis of pivotal study samples tested using fully validated immunogenicity assays, specifically the latest iteration of assays developed in the drug program. To include an integrated immunogenicity dataset from multiple clinical studies, for samples initially analyzed using an earlier set of assays, samples should be reanalyzed using the latest validated assays, or a bridging study may be performed that includes prior ADA positives, negatives and borderline samples. Strong concordance between methods may allow for pooling of results. Sponsors should justify the choice of samples that were reanalyzed. Data concordance for prior positives (strong and borderline) versus negatives should be evaluated.
As stated in the 2019 FDA immunogenicity guidance [52]:
Reproducibility is an important consideration if an assay will be run by two or more independent laboratories during a study, and a sponsor should establish the comparability of the data produced by each laboratory. Comparable assay performance, including sensitivity, drug tolerance, and precision, should be established between laboratories [52].
When it is planned to pool data from different methods and/or laboratories, ADA method cross-validation is needed to ensure that results from the different sources are sufficiently aligned. Lack of concordance prevents assessment of pooled data as this impacts the analysis of the ADA response prevalence/incidence, kinetics, and titer quartiles. Recent case studies were discussed to demonstrate different strategies to perform cross-validation studies. Evaluation of total agreement of antibody status results (negative/positive) provides a way to assess qualitative comparability. Results are recommended to match for 80–90% of assessments, however and this criterion should consider expected variability in the parent lab. As data can be summarized in a 2 × 2 contingency table, a statistical assessment such as the Cohen’s Kappa Test for the portion of agreement between categorical results can be used to determine the strength of agreement between the two methods.
Methods with a titer read out may also include an assessment of titer concordance (expect 80–90% match). Although, in general, titers within a single dilution are considered equivalent, this may be too stringent of a criterion to use for cross-validation between labs where titers within two dilutions (or titer change used to determine a boosted response) may be considered acceptable. When titers are interpolated from a dilution curve, minimum significant ratio (MSR) may be used to assess variability in titer results at the parent laboratory and this value can then be used to determine an acceptable titer range at the new laboratory (i.e., X% of titer results must be within the MSR of the original assay).
The impact of sample number was discussed as well as benefits of the use of “real” incurred samples versus spiked samples. Although it is preferred to use incurred samples, depending on the immunogenicity of the drug, it may be difficult to obtain a sufficient number of positive samples to cover a range of antibody levels equivalent to the low PC and higher. Preparation of spiked samples provides an alternative strategy to generate test samples and may provide some benefits to ensure samples with concentration levels below the low PC which may not consistently test positive in the assay are not included in the study. Refer to Figure 1 for a sample decision tree for the cross-validation of ADA assays.
Integrated Summary of Immunogenicity Harmonization
An ISI document is recommended in regulatory dossiers for biopharmaceutical product submissions. The overall aim is to provide a summary of the immunogenicity profile of a drug that was established during the product clinical development process and ultimately contribute to product approval and labeling. This involves a multidisciplinary approach including an immunogenicity risk assessment, Chemistry, Manufacturing and Controls considerations, the bioanalytical strategy, immunogenicity clinical sampling plan and evaluation of ADA and NAb impact on PK, clinically relevant PD biomarkers, efficacy, and safety parameters. Industry continues to assess approaches that efficiently comply with this recommendation. Regulatory input on this topic were provided in 2019 and 2020 [25,28], and the discussion continued in 2021, offering best practices for the immunogenicity assessment in support of submissions, providing the global bioanalytical community with the current industry/regulator consensus. Experience and lessons learned for the development of several ISIs submitted to health authorities were discussed.
The ISI helps establish the drug’s immunogenicity profile with a comprehensive, integrated evaluation of immunogenicity findings from clinical studies in the indication patient population. It is part of the product dossier for submissions. The ISI is generally viewed as a “living” integrated summary document that is used for information collection beginning early in product development (i.e., during research and candidate selection), and updated as the clinical program progresses through IND/Investigational Medicinal Product Dossier (IMPD) stages into BLA/Marketing Authorisation Application (MAA) submission and post-approval. For industry, it could be beneficial to use a folder approach, where relevant information and reports are stored with a developed ISI report template used for all programs. Consideration should be given to submitting a proposal for the content and analyses in the ISI as part of a Regulatory meeting.
Regulators recommend that sponsors divide the ISI into specific sections including background and overview, immunogenicity risk assessment, bioanalytical strategy for immunogenicity assessment, clinical study design and sampling strategies, clinical immunogenicity data analysis, literature review (optional, as needed), immunogenicity risk mitigation, overall conclusions, references, and appendices. Furthermore, it was recommended to provide a summary table as a snapshot of the relative risk while key risk factors are further detailed and discussed in the report. A visualization of the drug product has also proven helpful therefore a figure of the structural composition of the drug is suggested, with relevant chemical modifications or engineered sites properly identified. Product and quality attributes should describe the components in the final drug product; a table with drug product batches used in each of the clinical studies should be provided. Key patient and clinical trial factors that could influence immunogenicity should be included in the ISI. Clinical trial factors should include randomization, dose escalation, and treatment duration. Patient factors can be described in a table or figure of patient disposition across relevant clinical studies highlighting exceptions with enrolled patients.
The Bioanalytical Strategy section should be used as a summary for relevant bioanalytical methods, since details are provided in Module 2 of the common technical document (CTD) and validation reports in Module 5. A table should be included that highlights assays used for relevant data evaluated in the ISI (i.e., ADA and NAb assays, clinically relevant PD biomarkers and PK assays). Assay titles, validation report numbers, intended use, clinical studies supported, and relevant information should be provided. An additional table should specifically describe the immunogenicity assays including the method summary and an overview of the key assay performance parameters from the validation report. Hyperlinks to other documents should be used so that information is not repeated.
The Clinical Immunogenicity Data Analysis section should be completed in collaboration with statisticians, safety and clinical development teams. The recommendation was given to develop a Statistical Analysis Plan (SAP) for the ISI report. Populations should be primarily based on the safety database but may need further refinement as certain analysis will be directly from the clinical study. This keeps analyses focused, generates the required tables, listings and figures (TLFs) and helps align with safety pooled analyses. The consensus for suggested analyses populations include adult, pediatric, and overall (as appropriate). Immunogenicity data from studies could be pooled to provide a longitudinal integrated analysis, if appropriate and supported by the assays used.
The information in the Overall Conclusions section will be used for the Module 2 sections. The overall conclusions should be described succinctly, since a more detailed descriptive conclusion is at the end of each section of the ISI. Relevant risk evaluations and mitigation strategies should be included.
China NMPA Immunogenicity Guidance
The Chinese National Medical Products Administration recently issued its guideline on immunogenicity assessments for protein therapeutics [143] after similar documents were issued by EMA [53] and FDA [52]. There is not currently an official version of this document in English issued by NMPA but several versions are circulating within the industry translated by private companies. FDA, EMA and NMPA documents have similar wording and strategies for assessing and reporting immunogenicity of protein therapeutics. Similarities and differences from all three guidance documents regarding strategies as well as general guidance on assay development and validation have been previously discussed [144]. For example, in the NMPA guidance, it is stated that the approach for assessment should start from the preclinical phase and end with the clinical assessment, with a focus on cut point and sensitivity in the context of drug tolerance, selectivity and precision. Detailed instructions are provided on PCs, negative controls, assay parameters and evaluations; this includes precision criteria of <20% CV, which is more rigid than the criteria in the EMA guidance. There is an emphasis on cytokine measurement in immune responses and instructions are given for biosimilar immunogenicity assessments. There is also the need to include repeatability, robustness and stability during pivotal and post-marketing studies. The guidance for cut point assessment states that >15 individuals are needed for nonclinical and >50 individuals for clinical ADA assessment. Sensitivity requirements outline 250–500 ng/ml for nonclinical ADA and 100 ng/ml for clinical ADA evaluated in adequate drug tolerance, but multiple ways are allowed for calculating sensitivity. The minimum required dilution should not exceed 1:100.
The general process for a pre-approval inspection (PAI) begins with a notification of the inspection usually one week before the non-negotiable inspection date. The inspections start with an opening meeting, can continue with an on-site inspection which typically lasts between 1–3 days, followed by a close-out meeting. Responses to any findings should be prepared within 7 days, after which a final report is generated.
NMPA inspectors are more data-focused rather than process-focused. They inspect to ensure data integrity, accuracy and completeness. This is not to say that good quality management systems and processes are not important; on the contrary, good process drives performance. Site source document creation, maintenance, printing, copying, updating, and archiving can be inspected. Sample collection and handling inspections could include collection conditions, processing, extraction, storage, chain of custody and shipment. The storage, accountability, and traceability describing investigational medical product management is also a critical process that may be inspected. Training and equipment records may be considered.
For bioanalytical methods, important points of inspection could include shipment, storage and stability of drug materials; description of PK, ADA, and NAb methods; description of method development, validation and sample analysis processes; rationale for sample collection time points (PK/ADA); flow of bioanalytical data (instrument raw data to reported results to modeling); data storage, backup, and transfer; and, demonstration of the ability to reconstruct PK parameters with ∼10% of the PK data.
Recommendations were given regarding the need for validation of non-clinical ADA assays. Some companies are taking a “lean” approach for non-clinical GLP studies for low-risk molecules where the ADA assay is developed but generally not validated unless warranted by safety findings in the preclinical species (IC, hypersensitivity reaction, complement activation). If needed, validation is abbreviated (screening CP with n = 30 and FPR at 1%, drug tolerance, sensitivity and precision). The ADA assay is run in the screening format and results reported as positive/negative (with or without a SNR). If the intent is to explain or understand exposure (i.e., no impact on safety, PD), a generic ADA assay can be considered. Some companies conduct ADA investigation as an exploratory study. While ADA assays may be validated or FFP, some companies conduct sample analysis as a GLP study (include use of standard operating procedures, quality assurance, report, etc.). A validated method may be needed although NMPA regulators do have experience with FFP assays. It is recommended to consult with the agency given that there is no uniform approach followed.
It was agreed that these recommendations are not new but have been strictly enforced since 2019. General permits are needed for sample export (2–3 month delay) and additional shipments need supplemental permits (∼2 months or less). A PK study in the local population before Phase III studies is required. A pre-approval audit may also be performed.
Multi-Domain Biotherapeutics & Bispecific Antibody Immunogenicity
The advancement of bioengineering technologies in recent years has enabled development of complex protein therapeutics different from a conventional mAb format. Bispecific antibodies (BsAbs) are capable of binding to two different molecular antigens, or two different epitopes on the same antigen providing potential clinical benefits that traditional protein therapeutic formats could not offer. It takes more bioengineering modifications and complex manufacturing processes to generate BsAbs compared to the conventional mAb therapeutics. Therefore, BsAbs may potentially hold greater risk for immunogenicity and present unique challenges for immunogenicity analysis. Recommendation on BsAb immunogenicity and the value of applying integrated approaches for BsAb immunogenicity assessments were discussed based on new case studies, and newly acquired knowledge to update previous recommendations [12,28].
As a relatively new modality, development of BsAbs has a shorter history and may encounter unforeseen challenges, including bioanalytical ones. Well-established, general immunogenicity assessment strategies for the conventional mAbs may not be adequate for BsAbs. Customized bioanalytical strategies and methodologies should be planned for individual BsAb programs based on the target biology, mechanism of action, development phase and key reagent availability. Additionally, using a combination of tools should be considered for BsAb immunogenicity risk assessment. While the predictive tools have limitations to precisely envision the immunogenicity risks of a BsAb, during early development they do provide valuable information prior to in vivo data becoming available. There are limited publications and communications focusing on clinical experiences of immunogenicity results in BsAb programs. Therefore, it is premature to draw a general conclusion and categorize BsAbs as a product class as being highly immunogenic. It is recommended to share experiences from BsAb program development work to benefit the bioanalytical community.
Considerations were discussed for BsAb immunogenicity assessments, which are product-related (i.e., presence of immunogenic structural or linear epitopes). Critical assay reagents should be selected and characterized to include PC and cell line considerations for NAbs. ADA domain specificity should be evaluated since immune responses could inhibit the specific function of one domain while leaving the others intact, informing interpretation of other clinical readouts. Assay parameters should be established during validation within the relevant contexts (e.g., target interference; sensitivity and drug tolerance; threshold of each characterization assay).
In two case studies, integrated approaches were implemented to understand the characteristics of immunogenicity to BsAbs. In the first case study, a combination of an in vivo animal study, in silico analysis and in vitro T cell assay was used to evaluate the immunogenicity of anti-A/B and variant molecules. The anti-B Fab sequence was identified as a key contributing factor to high immunogenicity and there was no difference in anti-B positioning relative to the arm containing the knob or the whole mutations. Low immunogenicity prediction and in vivo results of anti-A/A suggested that the knob-in-hole platform is not necessarily immunogenic. Similar to the first case study, unexpected immunogenicity of LY34115244 was discussed in the second case study. In the LY3415244 (anti-TIM-3/PD-L1) Phase I study (Q2W IV flat doses from 3 to 70 mg), ADAs were detected in 12 of 12 patients and anaphylaxis was observed in 2 of 12 (17%) patients [145]. Pre-existing reactivity was characterized but indicated not to be predictive of the immunogenic responses in the clinic. Treatment duration correlated with a trend of increasing ADA responses. ADA responses had a negative impact on drug exposure. In both case studies, high immunogenicity was observed in the BsAbs but not in the corresponding parental molecules, which highlighted the higher immunogenic potentials in BsAbs.
Immunogenicity screening, confirmation and titer assays provide little information on ADA characteristics. A purpose-driven analysis of ADA isotype and specificity may be beneficial to understand immune response and their potential impact, using established methods like domain competition assays (DCA) and domain detection assays (DDA), where the biotherapeutic is enzymatically cleaved into individual domains, for example IgG or IgG-like modalities and fusion proteins [146]. Due to limited utility for smaller biotherapeutics, new approaches using protein engineering have also been more recently used [147].
The differentiation of ADA IgM and IgG isotypes may allow for conclusions on response progression and potential safety concerns which can justify placing effort into the development and application of such assays. The purpose of ADA isotyping is to detect hypersensitivity reactions, ICs, proof of antibody response (exclusion of soluble target), and time course of immune responses, leading to improved drug molecules. ADA responses are polyclonal and can be diverse, e.g., different Ig isotypes, directed against different epitopes [148–150]. There are multiple techniques for ADA isotyping including multiplexed LBA (e.g., Luminex xMAP Technology) [151], SquidLite Technology [152], electrochemiluminescence (MSD) [153,154], Genalyte Maverick System [31], and Immunocapture LCMS [155]. The PC is a critical reagent in this application (purified ADA chemically conjugated to antibodies of the respective isotype or recombinant ADA of the respective isotype) [153,156,157] and has to be rather sophisticated. A case study was discussed of ADAs against a targeted immunocytokine [158]. ADA isotyping showed 16 out of 18 patients were IgM positive, with 5 patients IgM-only positive and 11 patients with additional ADA IgG positives. Only ADA IgM responses were detected after 4 days and remained positive until Day 266 (last time point), but with a decreasing signal.
The development of an appropriately designed PK assay that is sensitive for ADA impact on relevant exposure could become an alternative strategy to understand part of the neutralizing potential of ADAs. Defining important criteria for PK assays becomes even more important for multi-functional therapeutics such as bispecific T cell engager molecules due to increased functional drug complexity. Deeper evaluation of critical reagents, formats and ADA positive controls are pre-requisites for MOA-based PK assay development and the generation of appropriate exposure data. Already in the preclinical space, the generation of relevant anti-idiotypic (anti-id) ADA PCs enables an accurate assessment of impact on exposure and an increased understanding of the potential ADA interference on drug target binding. Retrospectively proving the clinical validity of the assay performance might be a necessary step and another important criterion. Clinical proof can be assessed with additional advanced structural binding experiments of patient-derived ADAs to specific drug site domains. Identified patient ADA binding features can be used for selection of appropriate ADA PCs with similar binding features to confirm the clinically relevant impact of ADAs on exposure tested by the MOA-based PK assay. Comparable results of impact on exposure by patient-derived ADAs and selected ADA PCs suggests that observed reduction on exposure in patients is caused by the neutralizing potential of ADAs and allows correlation of ADA response and loss of exposure. Discussions provided important preclinical and clinical parameter criteria for the development of an appropriately designed PK assay for bispecific molecules to use as an alternative option towards understanding of neutralizing ADA impact on exposure. An appropriately designed PK assay is more informative than a total PK assay and, in consultation with health authorities, can be used in lieu of a NAb assay.
Biosimilar ADA Assay Validation & Harmonization
Health authority guidelines on immunogenicity assessment of biologics including expectations on ADA assays have evolved in the last few years [159]. In biosimilar development, specific requirements for immunogenicity assays need to be considered due to the comparative nature of clinical studies, i.e., head-to head evaluation of the immunogenic profile of a proposed biosimilar compared to a reference product. Often, a parallel study design is applied. In general, the use of the one assay approach to detect ADAs derived by the biosimilar candidate and its reference is accepted. This is mainly to enable a sensitive comparison of the immunogenicity of both products using one assay with the same characteristics compared to the application of different assays (one for the biosimilar and one for the reference) bringing additional variability to the analysis [160]. The one assay approach is preferred as it minimizes variability while determining relative immunogenicity. An additional benefit is that sample analysis can be blinded with a single data set by treatment group, minimizing discordant results. While ADAs against the biosimilar are readily detected by using one assay comprising the biosimilar as capture/detection reagent, a potential drawback may be a bias of greater sensitivity in detecting ADA against the biosimilar and the underestimation of ADA against the reference product. Therefore, the demonstration of assay suitability to detect anti-biosimilar and anti-reference product antibodies in the same manner is crucial in biosimilar ADA assay validation.
The one assay approach is supported by health authorities [52,53,161]. While the guidelines do not provide details on how to demonstrate suitability of the one assay approach, it has been discussed within the scientific community in recent publications [21,25,28,160,162]. A key experiment providing evidence for the suitability of one assay to detect ADAs towards the biosimilar candidate and reference is the evaluation of immunodepletion curves obtained by spiking either biosimilar candidate or reference to a fixed concentration of the PC. Similarity of the curves has been considered as the rationale to demonstrate appropriateness of the assay to detect both ADAs towards the proposed biosimilar and the reference product. However, acceptance criteria for the experiment have not been uniformly defined and are set arbitrarily. The same is true for the strategy on how to execute method comparability experiments, e.g., source of capture reagent, which fixed concentration of the ADA PC should be selected as well as the number of different concentration levels to be used in order to obtain sufficient data for a meaningful conclusion. Industry attendees suggested that the acceptance criteria are typically visibly overlapping curves and/or %CV of the mean signal from both drugs <20% at the majority of concentrations. Differences may be more prominent at higher ADA concentrations and, therefore currently three ADA concentrations (high, mid and low) are recommended [162]. A low ADA concentration close to the cut point may not provide meaningful data to evaluate immunodepletion, especially in the case of highly sensitive assays.
Another approach proposed was to perform immunodepletion analysis based on SNR instead of signal strength alone to identify the impact of the noise level on the analysis. As discussed, this also supports selection of appropriate ADA PCs and drug concentration levels as well as data interpretation.
An alternative for determining acceptance criteria based on a biostatistical approach was discussed, although current guidelines and publications do not provide recommendations for this approach. The prerequisite to apply a biostatistical approach to define margins and acceptance criteria is the understanding of the variability of the data and definition of the number of required repetitions. By applying a biostatistical approach, a high number of experiments (≥12) would most likely be required [163–165]. It is questionable if such a detailed characterization of this validation parameter provides additional key insights into assay performance and is therefore not recommended. Selecting appropriate ADA PCs and corresponding drug concentration levels already ensures a sensitive evaluation of curve similarity and the number of repetitions of such experiments needs to be decided based on assay variability on a case by case basis.
Recommendations were discussed for ADA PC selection. The ADA positive control is a surrogate for the immune responses which may occur in humans, mainly serving to characterize assay performance. The nature, source and type of the positive control can be diverse. In assay validation, ADA PCs are often generated by hyperimmunization of animals. Polyclonal ADA PCs may reflect the immune response in humans to a greater extent than using in vitro manufactured mAbs. To generate a relevant positive control, immunization with the drug without human IgG Fc or purification for Fab or F(ab’)2 should be considered.
Anti-biosimilar and anti-reference antibodies are adequate PCs and either can be used for the one assay approach considering detection of ADAs is independent of ADA PC. It is more important to follow the conservative approach using the biosimilar for the capture and detection step. Positive controls, independent from which antigen was used for immunization, are surrogates and therefore have some limitations. The characteristics of the PCs are more dependent on individual responses of the immunized animal compared to differences derived from immunization with the biosimilar or the reference therefore consensus was reached that evaluating positive controls derived by immunization with biosimilar and reference are not considered to add value to assay validation but could provide increased confidence in results.
The demonstration of equivalent drug tolerance is key for applying the one assay approach. High drug tolerance at low ADA positive concentrations enables sensitive immunogenicity assessment. If the required drug tolerance could not be reached for all time points during a clinical study, an adequate sampling schedule must be implemented. Equivalent drug tolerance for biosimilar and reference should be demonstrated to allow a meaningful comparison of the immunogenic profiles. Drug tolerance is generally tested in the screening assay but should also be considered in the confirmatory assay [162].
Additional experiments to assess confirmatory assay performance using individual samples as recommended by Civoli et al. [160] were discussed and it was not considered mandatory to include them to test the suitability of the one assay approach. It is suggested that positive control spiked samples (10–15) should be tested in the presence of excess of biosimilar or reference on the same plate at PC concentrations giving SNR ∼1.5 – 2. The percent inhibition of the assay signal of the drug-spiked samples relative to the unspiked samples should then be determined separately and compared between the two drugs. Since the appropriateness of the one assay is already addressed with the immunodepletion analysis and drug tolerance experiments, the described additional testing is not considered to add further relevant information to the assay performance.
RECOMMENDATIONS
Below is a summary of the recommendations made during the 15th WRIB:
NAb Assays – Drug & Target Interference
- NAb assays (both cell based and ligand binding NAb assays) are susceptible to false positive or negative assay results in the presence of multimeric soluble targets. At high concentrations for monomeric targets, this may be associated with either homodimerization of the target or aggregation during sample processing.
- Prior exposure to biologics binding the same target increases the potential for prior drug interference. This needs to be investigated and addressed during bioanalytical assay development and validation.
- High concentrations of soluble target can be mitigated by selecting screening assay conditions that preserve the positive assay signal to allow setting of a scientifically sound cut point for the confirmation step, or by adding a soluble target depletion step.
- The fundamental parameters for ADA and NAb assays validation include (1) cut point; (2) sensitivity and drug tolerance; (3) specificity and selectivity; (4) precision; (5) reproducibility when relevant; (6) robustness of some assay features; and (7) in-use stability of critical reagents [52].
- Sponsors should summarize the efforts conducted and focus on a FFP philosophy when developing and validating assays.
- The USP guidance [110] is suggested as another possible resource until a NAb assay validation guidance is available.
- Cell-based NAb assay is generally less sensitive compared to competitive LBA. To improve drug tolerance, variations of acid dissociation and interfering drug or target immunodepletion are more common. Regardless of method, it is key to demonstrate adequate diligence has been exercised within the limitation of the available PC reagents during assay development and assay validation.
- A balance between sensitivity and drug tolerance is needed such that an assay has adequate sensitivity in samples with anticipated trough levels of drug. Regulators may ask for “wash out” samples evaluating patient immunogenicity response to the drug.
- An integrated approach can also be considered for “totality of the evidence,” in combination with other available data.
- The integrated data approach evaluates impact on exposure, efficacy and safety by correlating levels and persistence of binding ADA to PK, PD, biomarkers and safety events and can also include relevant in vitro NAb assay data.
- Where there is a highly sensitive PD marker or an appropriately designed PK assay (or both) that generate data that inform clinical activity, it may be possible to use these in lieu of a NAb assay. This determination should be done in consultation with the Agency.
- Consider if the PK assay can measure functional drug in the presence of ADA and if it is inhibited by ADA.
- Regulators recommend including free ligand measurement and clinical effect in the integrated data approach when relevant.
- Regulators request the inclusion of NAb data to enable them to build a picture of what is happening in a patient. For low-risk molecules, if the immune response does not have an impact on safety, not performing a NAb assay may be discussed and negotiated with the agency on a case-by-case basis; gaps will result in requests for data pre-approval or in a PMC.
ADA Cut Points – Appropriateness & False Positive Rates
- It is critical to understand the relationship between cut points and sensitivity/specificity in addition to false positivity.
- Alternative approaches for cut point determination were discussed, such as using assay variability (minimum/precision cut point), achieving acceptable assay sensitivity (e.g., ≤100 ng/ml), and using the assay specificity curve and visualizing/gating the data. Discussion with regulatory agency is recommended for implementing alternative approaches.
- Alternative cut point data transformations might improve the FPR.
- Consider reporting ADA incidence for placebo on the label.
- Industry suggested using FPR <1% for drugs with low immunogenicity (e.g., human mAbs). However, this was not supported by the regulators.
- There can be instances where mathematical modelling for cut point assessment can be helpful (e.g., pediatric populations, rare diseases), however, there is insufficient data for using this method in a more general way.
Circulating Immune Complexes (CIC) – ADA/Drug Complexes
- Research study discussed supports that serum was a suitable matrix for IC analysis.
- Increased clearance of ICs is not an ADA property, but rather every ADA (monoclonal and polyclonal) can have clearing potential when the formed ICs have a critical size (dependent on ratio and concentration of drug and ADA) in combination with the structure of the formed ICs.
- IC formation results in enhanced proteolysis and renal excretion.
ADA Assay Comparability
- Regulators do not currently recommend performing ADA assay comparability using platform agnostic methods to compare immunogenicity rates across different drug programs from different sponsors.
- When different sets of assays are used within the same immunogenicity program, data supporting labelling should be based on the analysis of pivotal study samples tested using fully validated immunogenicity assays (i.e., the latest iteration of assays).
- To include an integrated immunogenicity dataset generated from multiple methods and/or laboratories:○A bridging study should be performed that includes prior ADA positives, negatives and borderline samples.○Justification should be provided for the choice of the reanalyzed samples.○Data concordance for prior positives (strong and borderline) versus negatives should be evaluated.○When acceptable concordance cannot be established, samples initially analyzed using an earlier set of assays should be reanalyzed using the latest validated assays.
- A statistical assessment such as the Cohen’s Kappa test can be used to determine the strength of qualitative agreement (positive/negative) between two ADA methods.
- Methods with a titer read out may also include an assessment of titer concordance (expect 80–90% match). Use of a statistical assessment, MSR, to assess variability in titer results at the parent laboratory can be used to determine an acceptable titer range at the new laboratory when titers are interpolated from a dilution curve.
- It is preferred to use a sufficient number of positive incurred samples to cover a range of antibody levels equivalent to the low PC and higher. Preparation of spiked samples provides an alternative strategy to generate test samples if incurred samples are not available or if there are not enough positive study samples. It is important to evaluate both positive and negative samples, particularly when a statistical assessment will be performed.
- Comparability criteria should consider expected variability of the parent method.
- Use of the amount of CP-ARC as a defined, PC-independent assay parameter that enables comparison of technical assay performance was proposed.○The amount of ARC needed is the signal generating ADA/analyte reagent complex required to give an assay signal at the cut point.○CP-ARC is a constant assay performance parameter that defines the assay sensitivity and drug tolerance○The interaction between ADA and drug and the interaction between ADA and labelled drug/reagents should be identical.○Assay conditions and reagents should be appropriate to form defined ARCs.○Interactions between ADA, PC and reagents need to reach equilibrium.○Correct concentrations of (active/binding competent) reagents and PCs are required.○Additional data, examples and experience are needed to further support the use of CP-ARC in ADA assay cross-validation.
Integrated Summary of Immunogenicity Harmonization
- Information for the ISI should be collected beginning early in product development (i.e., during research and candidate selection) and updated as the clinical program progresses through IND/IMPD stages into BLA/MAA submission and post-approval.
- It is beneficial to use a folder approach, where relevant information and reports are stored with a developed report template used for all programs.
- Consideration should be given to submitting a proposal for the content and analyses in the ISI as part of a Regulatory meeting.
- Sections of an ISI:○background and overview;○immunogenicity risk assessment;○bioanalytical strategy for immunogenicity assessment;○clinical study design and sampling strategies;○clinical immunogenicity data analysis;○literature review (optional, as needed);○immunogenicity risk mitigation;○overall conclusions;○references;○appendices.
- It was recommended to provide a summary table as a snapshot of the relative immunogenicity risk including assessing product and quality related factors as well as clinical-related (posology) and patient factors. Relevant aspects can be further described.
- A figure of the structural composition of the drug is suggested, with relevant chemical modifications or engineered sites properly identified.
- The components in the final drug product should be noted and a table with drug product batches used in each of the clinical studies should be provided.
- Clinical trial factors should be described including randomization, dose escalation, and treatment duration.
- Patient factors can be described in a table or figure of patient disposition across relevant clinical studies with exceptions in enrolled patients highlighted.
- The bioanalytical strategy section should be used as a summary for relevant bioanalytical methods, since details are provided in Module 2 of the CTD and validation reports in Module 5.
- Bioanalytical Strategy section:○A table should provide assays used for relevant data evaluated in the ISI (i.e., ADA and NAb assays, clinically relevant PD biomarkers and PK assays).○Assay title, validation report number, intended use, clinical studies supported, and relevant information should be provided.○Another table should include the method summary and an overview of the key assay parameters from the validation report.○Hyperlinks to other documents should be used so that information is not repeated.
- Clinical Immunogenicity Data Analysis section:○complete in collaboration with statisticians, safety and clinical development teams.○develop a SAP for the ISI report to keep analyses focused, generate required TLFs and help align with safety pooled analyses.○The consensus for suggested analyses populations include adult, pediatric, and overall (as appropriate) with longitudinal analyses populations data pooled, if appropriate and supported by assays used.
- Overall Conclusions section:○will be used for the Module 2 sections○should be described succinctly○relevant risk evaluations and mitigation strategies should be included
China NMPA Immunogenicity Guidance
- Some NMPA validation parameters for ADA and NAb assays differ from the FDA and EMA guidance/guideline:○assessment should start from the preclinical phase and end with the clinical assessment;○focus on cut point and sensitivity in the context of drug tolerance, selectivity and precision;○detailed instructions are provided on PC, NC, assay parameters and evaluations;○precision criteria <20% CV;○emphasis on cytokine measurement in immune responses;○instructions are given for biosimilar immunogenicity assessments;○need to include repeatability, robustness and stability during pivotal and post-marketing studies.
- Cut point assessment:○>15 individuals are needed for nonclinical ADA assessment;○>50 individuals for clinical ADA assessment.
- Sensitivity:○250–500 ng/ml for nonclinical ADA;○100 ng/ml for clinical ADA;○evaluated in adequate drug tolerance.
- Minimum required dilution should not exceed 1:100
- General process for a pre-approval inspection (PAI):○notification of the inspection usually one week before the non-negotiable inspection date;○opening meeting;○on-site inspection (1–3 days);○close-out meeting;○responses to any findings prepared within 7 days;○final report is generated.
- Validation of non-clinical ADA assays for low-risk molecules:○Generally not validated unless warranted by safety findings in the preclinical species.○If needed, validation is abbreviated (SCP with n = 30 and FPR at 1%, drug tolerance, sensitivity and precision).○assay is run in the screen format and results reported as positive/negative (with or without a SNR).○If the intent is to explain or understand exposure (i.e., no impact on safety, PD), a generic ADA assay can be considered.○It is recommended to consult with the agency given that there is no uniform approach followed.
- General permits are needed for sample export (2–3 month delay) and additional shipments need supplemental permits (∼2 months or less).
- A PK study in the local population before Phase 3 studies is required.
Multi-Domain Biotherapeutic & Bispecific Antibody Immunogenicity
- Customized bioanalytical strategies and methodologies should be planned for individual BsAb programs based on the target biology, mechanism of action, development phase and key reagent availability.
- Predictive tools have limitations to precisely envision the immunogenicity risks of a BsAb, but they do provide valuable information during early development prior to clinical data becoming available.
- It is premature to draw a general conclusion and categorize BsAbs as being highly immunogenic.
- PCs and cell lines for NAb assays have to be selected and characterized carefully.
- ADA domain specificity should be evaluated case by case.
- Assay validation parameters should be established within the relevant contexts.
- An integrated approach should be considered to understand the characteristics of immunogenicity to a BsAb.
- A purpose-driven analysis of ADA isotype and specificity might be beneficial to understand immune response and their potential impact.
- The differentiation of ADA IgM and IgG isotypes might allow conclusions on response progression and potential safety concerns which justify placing effort into development and application of DCA and DDA assays.
- PK assay development of bispecific molecules is based on MOA and correlates with ADA impact on exposure using available critical ADA PCs.
- Identification of an anti-id reagent as a valid ADA surrogate with similar binding features to clinical ADAs is suggested to evaluate the relationship of ADA impact on exposure.
- Suitable ADA PCs against high-risk domains (e.g., domain with endogenous counterpart) is important.
Biosimilar ADA Assay Validation & Harmonization
- A key experiment providing evidence for the suitability of the one assay to detect ADAs towards the biosimilar candidate and reference is the evaluation the similarity of immunodepletion curves.
- Traditional approach for assessing immunodepletion:○Immunodepletion curves are prepared by spiking matrix with a known concentration of ADA and increasing concentrations of reference and biosimilar.○The acceptance criteria is typically that the curves should be visibly overlapping.○Three ADA concentrations are recommended.
- Another approach proposed by the industry was to perform immunodepletion analysis based on SNR instead of signal strength alone.
- Low ADA concentration levels may not allow a meaningful evaluation of immunodepletion curve analysis and, especially when highly sensitive ADA methods are applied, the use of very low ADA concentration is not recommended.
- An alternative for determining acceptance criteria may be the use of a biostatistical approach but was considered not being required in general. The prerequisite is the understanding of the variability of the data and the definition of the number of required repetitions which needs to be decided case by case.
- ADA PC selection:○Polyclonal ADA PCs may reflect the immune response in humans to a greater extent than using in vitro manufactured mAb and should be considered for assay validation.○Immunization with the drug without human IgGFc or purification for Fab or F(ab’)2 should be considered.○Anti-biosimilar and anti-reference antibodies are adequate PCs, and either can be used.
- The conservative approach considers the use of the biosimilar for the capture and detection step.
- Comparing different positive controls derived by immunization with biosimilar and reference are not considered to add value to assay validation.
- Similar or equivalent drug tolerance for biosimilar and reference should be demonstrated to allow a meaningful comparison of the immunogenic profiles.
- If the required drug tolerance could not be reached for all time points during a clinical study, an adequate sampling schedule must be implemented.
- Drug tolerance is generally tested in the screening assay but should also be considered in the confirmatory assay.
- Additional experiments to assess confirmatory assay performance using individual samples are not considered mandatory to test the suitability of the one assay approach.
Acknowledgements
- US FDA, Europe EMA, UK MHRA, Austria AGES, Norway NoMA, Brazil ANVISA, Health Canada, Japan MHLW and WHO for supporting this workshop.
- Dr. Eugene Ciccimaro (BMS), Dr. Anna Edmison (Health Canada), Dr. Fabio Garofolo (BRI), Dr. Swati Gupta (AbbVie), Dr. Shannon Harris (HilleVax), Dr. Carrie Hendricks (Sanofi), Dr. Sarah Hersey (BMS), Dr. Steve Keller (AbbVie), Dr. Lina Loo (Pfizer), Dr. Mark Ma (Alexion), Dr. Joel Mathews (Ionis), Dr. Meena (Stoke), Dr. Manoj Rajadhyaksha (Alexion), Dr. Ragu Ramanathan (Vertex), Dr. Susan Spitz (Incyte), Dr. Dian Su (Mersana), Dr. Matt Szapacs (Abbvie), Dr. Albert Torri (Regeneron), Dr. Jian Wang (Crinetics), Drs. Jan Welink (EU EMA), Dr. Yuling Wu (AstraZeneca) for chairing the workshop and the White Paper discussions.
- All the workshop attendees and members of the Global Bioanalytical Community who have sent comments and suggestions to the workshop to complete this White Paper.
- Future Science Group as a trusted partner
Financial & competing interests disclosure
The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.
No writing assistance was utilized in the production of this manuscript.