16^th GCC Closed Forum: ICH M10 implementation; NGS, qPCR/dPCR, flow cytometry validation; tissue biomarkers; IS response; immunogenicity harmonization; bioanalytical industry status

Abstract

The 16th GCC Closed Forum was held in Orlando, FL, USA, on 23 June 2023. Representatives from international bioanalytical Contract Research Organizations were in attendance in order to discuss scientific and regulatory issues specific to bioanalysis. The issues discussed at the meeting included: IS response, flow cytometry, changes to the bioanalytical industry, NGS assays, biomarker assay for tissues, dPCR validation, immunogenicity harmonization and ICH M10 implementation. Conclusions and consensus from discussions of these topics are included in this article.

1. Background

The Global CRO Council in Bioanalysis (GCC) is an independent global consortium created in 2010 bringing together CRO leaders to discuss various topics and challenges on scientific and regulatory issues related to bioanalysis while working with many different sponsors, vendors and regulatory agencies [1]. Since its formation, GCC has held regular meetings and published conference reports to share discussions and opinions [2–10]. White Papers on specific topics of widespread interest in bioanalysis and unified recommendations have also been published and were well received by the global bioanalytical community [11–20].

2. Introduction

The 16th Closed Forum of the Global Contract Research Organization (CRO) Council in Bioanalysis (GCC) was held on 23 June 2023 in Orlando, FL, USA.

Eight topics were discussed at the 16th Closed Forum; each included results of a survey designed by topic leaders which was circulated to GCC member representatives and compiled for presentation.

The eight topics discussions and leaders were:

• 1.What's the status of the ICH M10 implementation after a year from its issuing? (Elizabeth Hyer and Mark O'Dell)
• 2.Recommendations for validation of NGS assays (David Willoughby)
• 3.Does dPCR validation require independent recommendations from qPCR? (Amanda Hays, Amy Lavelle and Manisha Diaz)
• 4.Recommendations on practical considerations of Flow Cytometry validation (Shabnam Tangri and Naveen Dakappagan)
• 5.Biomarker assays for tissues: matrix, long-term stability (LTS) and assay validation (Moucun Yuan and Aihua Liu)
• 6.Evaluation of internal standard response (Jennifer Zimmer and Shane Karnik)
• 7.Current disagreements in immunogenicity harmonization (Todd Lester)
• 8.Changes to the bioanalytical industry – staffing, supply chain, capacity/lead time and costs (Eric Thomas)

This report summarizes survey results and discussions from the meeting for all topics.

3. Discussion topics

3.1. What's the status of the ICH M10 implementation after a year from its issuing?

In July 2022, the harmonized ICH M10 guidelines were published with the intention of providing recommendations for the validation of bioanalytical assays for chemical and biological drug quantification and their application in the analysis of study samples [21]. Since then, it has been adopted by the US FDA and EMA effective from November 2022 and January 2023, respectively. To analyze the implementation of the ICH M10 among CROs a year on from its publication, a survey of 29 questions was conducted with 28 GCC company's representatives' part-taking in the discussion.

General implementation aspects were first analyzed. In terms of the direct effect of the ICH M10 guidelines, roughly half of the respondents stated they saw no changes, and over a quarter felt it improved their projects, while 17.86% felt it created new challenges. The timing of the updates with respect to changes in experimentation procedures and formats for alignment with ICH M10 guidance for validation policies, validation report templates, sample analysis policies and sample analysis report templates followed a similar pattern with most (roughly 30% for all) implementing updates effective after adoption by the EMA (January 2023) and a similar proportion (20–30%) implementing after adoption by the US FDA (November 2022).

For both comparative bioavailability (BA)/ bioequivalence (BE) studies and other (non-BA/BE) sample analysis studies, the majority reported no additional time being required for reporting, with roughly 40% for both indicating a few extra days were needed under the ICH M10 guidance. Only for BA/BE studies did 8% of the respondents indicate an additional week or more being required for reporting. Although the ICH M10 guidance states that biomarkers are out of scope, it was interesting to note that 20.83% used Section 7.1 of the guidance for biomarker method validations at present. The majority (58.33%) follow the US FDA BMV guidance [22], with the remaining 20.83% stating to use a combination of resources. There could be an opportunity for the harmonization of the BMV and other resource documents used by industry. However, since biomarkers should ideally be fit-for-purpose, and any guidance would need to be specific for different uses, discussions emphasized such efforts to standardize practice would be welcomed.

Just over half of the respondents were routinely including over-range dilution QCs in matrix stability testing for all methods, while 32% were only using them for reasons such as: clinical sample concentrations being considerable over ULOQs, at sponsors requests, and some stating they are in the process of establishing routines. When study samples are observed to be above the ULOQ during a non-clinical study, over three quarters of the respondents stated they almost never determine if a new or modified calibration range is required and just include over-range QCs for study sample acceptance, promoting the session to raise the question of whether changes to the regulation should be suggested regarding the calibration range. Almost 20% stated they only determine if a new or modified calibration range is required if a certain percentage of samples are above the ULOQ (e.g. over 20 or 30% samples require dilution). When study samples are observed to be above the ULOQ during human trial studies however, there was a nearly-equal split of respondents PK between those who do not modify the range and those who do modify the range if a certain percentage of samples are above the ULOQ or certain criteria were not met, for example, examining PK profiles by each case to adjust the range if fall within or beyond the high end of the existing range. In both cases, consultation with sponsors was important. When exploring how respondents were selecting the regression model for a bioanalytical method, there was a roughly 50/50 split of those determining curve selection and documenting it during method development, with no further evaluation being performed within the validation, and those determining curve selection in method development, then performing a formal evaluation and documenting it during validation.

ICH M10 updates with relation to chromatography was then explored. When questioned about retrospective updates to previously validated LCMS methods to align with ICH M10 guidance (i.e., performing additional laboratory assessments), 37.5% respondents indicated they were doing additional experiments, for methods meeting certain criteria such as very old methods being updated to meet current standards in new projects, add missing sections, at the request of sponsors, and for updating drug–drug interaction assays validated pre-ICH M10.

The ICH M10 does not require a Certificate of Analysis (CoA) for IS as long as the suitability for use is demonstrated. The survey examined how laboratories were handling expiries and retesting dates for IS reference materials. It was interesting to note a total of 60.87% respondents were assigning the expiry/retesting date per the CoA and if there is none listed, were requesting data from the sponsor/vendor or assigning a standard expiry for a pre-set time frame. 17.39% were assigning an expiry/ retest date of ‘not applicable’ for IS reference material, and the remaining were following other routes. Concerns were expressed during the discussions regarding assigning an arbitrary expiry date which may not actually be demonstrating stability, especially when regulators often look at variability of curve slope. When looking at how laboratories were handling correction factors for IS reference material, the majority were assigning the correction factor as accurately as possible, just like reference material for an analyte of interest, while around a fifth were not assigning a correction factor. Concerns were raised in the discussions regarding if one does not correct for purity/potency, they could have significant lot-to-lot variation in IS responses, which will have implications on the method.

Looking at the practical aspects of implementation, no change in timing or only a few additional days were noted to be required for LCMS validation work under the ICH M10 guidance, while the expense for updating previously validated, sponsor proprietary, LCMS methods to meet the standards were predominantly handled by setting prices based on expected amount of work required to update the methods (65.22%). Challenges were however noted when sponsors felt the changes were unnecessary. It is also noteworthy that 30.43% stated they were not updating previously validated LCMS methods.

The regularity of matrix effects evaluation in relevant patient populations or special populations for LCMS methods appeared to vary. Around a quarter of the respondents stated that they rarely evaluated ‘non-standard’ matrix effects, while a fifth stated they evaluated it routinely only for hepatic and renal impairment studies, and a further 34.15% were evaluating routinely for all special populations. With high dependency on the patient population, it was agreed how these decisions on whether to evaluate matrix effects were reached must be discussed further for future guidance and harmonization.

For plasma LCMS, the most popular method to test for stability in whole blood was to spike whole blood with the analyte(s) of interest at two or more concentrations and process to plasma to use the plasma method for analysis (60.87%), and a further quarter of the respondents were spiking at a single concentration. Discussions noted there have been arguments for the use of exploratory whole blood method for this purpose.

Further exploring practice alignment with guidance updates, during the assessment of accuracy and precision in validation of an LCMS method, the QCs were found to most typically be analyzed fresh for one assessment then frozen for subsequent assessments (72.73%). With respect to selection of the regression model for an LCMS method, almost half of the respondents were processing the data using multiple regression models and selecting the one with the best QC accuracy (‘test and fit’), with roughly 40% defaulting to a 1/x2 linear model with no further documented assessment. Participants discussed whether or not the latter, if documented within a formal policy, aligns with the requirements of ICH M10 or if the guidance is specifically calling for a positive assessment of regression method.

Finally, LBAs were explored. With the ICH M10 noting the potential for singlicate analysis for LBAs, there was an equal split of respondents pursuing the option, and those who were not looking into it at this time (38.10% each), with about a quarter unsure about the possibilities. The responses were discussed to be dependent on company preferences, sponsor preferences and whether the studies were non GLP and so on, with some expressing interest in considering use in the future with the responses seen in the present survey. For those proceeding with singlicate analysis for LBAs, generally taking a retrospective approach, some noted additional validation experiments are performed to verify singlicate approach is appropriate (35%), while a smaller portion stated they used current validation data to determine the same (15%). With the ICH M10 noting stability should be assessed at every step of sample processing and accessioning, for LBAs, this would be around sample processing at time of analysis at the bench and does not include whole blood stability to be done every time, and therefore should be evaluated on a case-by-case basis. It was interesting to note while roughly half of the respondents agreed this to be the case, 28.57% disagreed stating whole blood stability should be included as part of stability assessments. It was noted that this was in fact questioned during remote audits and would also depend on factors such as the needs of sponsors, length of time at bench, whether offsite or onsite and so on. The need for whole blood stability testing was found to be evaluated through early contract discussions when scoping out the project by a third of respondents, included within current stability experiments by another third, while the remaining third stated they never performed whole blood stability for LBAs. The use of surrogate matrix application was another note from the ICH M10, which was found to be adopted by 57.14% of the respondents for LBA and of those using it, the majority were applying it for hard to obtain or rare matrices only (61.90%), or rare matrices only (14.29%).

The need for a well-defined lifecycle management for critical reagents is another requirement from the ICH M10, to help ensure consistency in assay performance. It is noteworthy that majority of respondents were already implementing lifecycle management for critical reagents (42.86%) or were in the process of implementing it (38.10%), with a small fraction stating they do not plan on doing so as their established SOPs already covered this. To finish, the default selection for curve fitting for an LBA was captured in an open question where the responses indicated most companies were complying with the ICH M10 guidance on the matter which states that the relationship between response and concentration for a calibration curve is most often fitted by a 4- or 5-parameter logistic model if there are data points near the lower and upper asymptotes (and other models should be suitably justified).

3.2. Recommendations for validation of NGS assays

Next-generation sequencing (NGS) is a powerful genomic analysis method, which determines the sequence of DNA or RNA to study genetic variations associated with diseases and other biological trends. The high-throughput and accuracy of NGS makes it ideal for targeted sequencing providing comprehensive genomic coverage, enabling applications in both genomic and clinical research.

This session focused on validation and use of NGS in regulated drug-development settings discussing the results of the 12-question survey, which was answered by 20 respondents. With 30% of respondents indicating their CRO is not currently involved in using NGS assays to support gene therapy, cell therapy or vaccine studies.

The main use of NGS was indicated to be for clinical monitoring, then discovery work, followed by gene editing risk assessments. Despite a selection of available guidelines [23–29] for NGS assay development and validation, for which the session leader presented performance metrics recommended for assessment highlighting common crossovers, it was clear the vast majority were not following established guidelines. The majority (90.91%) of the respondents stated they used in-house SOPs, while 45.45% indicated they used White Papers, with follow-up discussions indicating that some used PCR guidelines instead with known criteria, which was easier to follow. It was agreed there is value in establishing a set of best practices in NGS assay sample analysis and validation to serve as a source of guidance specific for CROs, with a unanimous vote from the respondents, although the difficulty to gather the various criteria into one document was acknowledged.

Further development and clarification of guidelines was voted to be required by 91.67% of the respondents when asked whether they thought the guidelines and regulations for performance of NGS assays for GLP pre-clinical work and clinical trials were well defined. The discussions emphasized they were not designed for GLP but for clinical studies, with a more clear, concise, and simplified documents required for fast-paced bioanalytical CRO laboratories.

It was evident that majority of respondents were not using the AMP guidelines [23] for NGS with over 80% indicating as so when questioned specifically on the guidance, about whether they felt the AMP level of validation for a research end-point assay was necessary and also about its use for inclusion/exclusion and patient stratification. Discussions highlighted that it wouldn't make sense to have a single validation for a variety of applications. A similar response was observed when asked if such documents leave grey areas around validation (such as in the exact determination of false-positive rates in NGS).

Using established standards, or a variety of samples with specific mutations or gene fusions were the approaches voted most popular when determining false-positive rates in NGS. All respondents were including software pipeline for genetic alterations calls in their guidelines. Some of the key factors considered as a CRO when developing NGS technologies to support sponsors working in cell or gene therapy, or vaccines were suggested to be the requirement of harmonized practices and guidelines specific for CROs working in bioanalysis; accuracy, streamlined and reliable pipelines, sponsor requirements; and knowledge of intended use.

It seems that NGS is expanding in the CRO business but right now not many CROs are directly working with NGS. Nonetheless, according to the survey several GCC members stated that they (or another decision-making member of their company) would be interested in contributing to the harmonization of NGS assays validation among CROs in a White Paper and the facilitator closed the session stating they would be open to starting these conversations.

3.3. Does dPCR validation require independent recommendations from qPCR? Follow up on the recommendation paper on qPCR/dPCR assay validation

The use of quantitative PCR (qPCR) and digital PCR (dPCR)/droplet digital PCR (ddPCR) in bioanalytical laboratories have increased as a result of the surge in gene therapy, cell therapy and vaccine research, with CROs undertaking development and validation of qPCR and dPCR assays. With limited guidance for the use of qPCR, and absence of regulatory guidelines for the use of dPCR to support regulated bioanalysis, the GCC issued a White Paper titled ‘Recommendations on qPCR/ddPCR assay validation by GCC’ [20] in 2022 to provide unified, focused and clear recommendations from the GCC on development and validation of qPCR/dPCR assays in bioanalysis to facilitate interactions with sponsors and regulators, and for the development of harmonized SOPs for qPCR/dPCR among CROs. These were based off survey results and expansion of prior recommendations to reach consensus GCC recommendations.

The present survey aimed to explore whether digital PCR (dPCR) validation requires independent recommendations from qPCR and consisted of 19 questions of which the key observations were discussed during the session. Of the 19 respondents taking part, 17 were currently using dPCR assays all of whom were using ddPCR, of whom 41.18% were using separate SOPs for dPCR as opposed to qPCR validations. While roughly 30% were evaluating the same performance characteristics in dPCR validations as in qPCR validations, the remaining were following recommendations from the GCC Harmonization White Paper.

The use of different parameters for dPCR validation studies were then explored in detail. While the GCC consensus stated no calibrators are required during sample analysis while a set of calibrators are required during validation, it was interesting to note that 35.29% were not using calibrators in dPCR assays with discussions revealing, for example, some use QCs of the standard curve. Roughly half of the respondents (47.06%) were using the acceptance criteria for precision and accuracy for dPCR at LLOQ as suggested in the GCC Harmonization White Paper. 58.82% were following the recommendations set forth in the GCC Harmonization White Paper for assessing linearity in dPCR validations, even if linearity wasn't considered as important in dPCR. Similarly, roughly half of the respondents were determining LOD and LLOQ for their sensitivity assessments as recommended in the White paper (47.06%), and 64.71% were following the recommendations for stability testing. Selectivity and specificity were being assessed as per GCC recommendations by 70.59% of the respondents.

The final set of questions examined features specific to ddPCR validation. Both manual and automated thresholds were being used in data analysis for dPCR assays with a roughly 50/50 split, with benefits of automated threshold discussed, including consistent throughput and removal of operator bias. There were discrepancies in the use of methods to compensate for partial volume variability with an almost equal split of respondents using them versus not using them. A similar pattern of a roughly 50/50 split was observed in pre-processing of dPCR validation samples (steps upstream of the dPCR workflow) compared with qPCR validation samples. Further variations were reflected in the response to whether respondents considered false amplification signals as acceptable in dPCR, with 41.18% saying they always did, 29.41% saying they did not, and 29.41% saying yes but depending on variable factors. This effectively means while some laboratories having false amplifications as acceptable, or acceptable under certain circumstances, to others it was unacceptable.

With a high level of discrepancies in validation assessments and use of guidance evident from the survey results and discussions, the session concluded with the call for a follow up White Paper on ddPCR based on the dissemination of GCC consensus and relevant case studies. This appeared to be supported by 88.24% of the respondents registering their interest in contributing to the efforts towards harmonization of dPCR assays validation among CROs.

3.4. Recommendations on practical considerations of flow cytometry validation

Flow cytometry is a versatile cellular analysis tool with the ability to conduct multiparametric measurements with high sensitivity and specificity. It is used in various stages of drug development and requires assays to be performed in a variety of regulated environments. Its applications continue to expand, alongside the need to address associated challenges and standardization of laboratory practices. This session focused on flow cytometry method validation, discussing the challenges and common solutions that balance regulatory requirements with laboratory practices. A total of 25 members responded to a survey consisting of 21 questions, of which 84% stated their CRO was currently involved in using flow cytometry assays and went on to respond to the remaining survey. Owing to the length of the survey, during the meeting the topic leaders focused the discussions on questions with the highest discordance.

Two-thirds of respondents indicated that their laboratory did not follow the Clinical and Laboratory Standards Institute guidelines for flow cytometry ‘H62 - Validation of Assays Performed by Flow Cytometry’ [30], for which reasons provided included the use of existing best industry practices, in-house SOPs, and preference for guidelines supported by regulatory bodies, and those to be published by the US FDA and EMA. There was some hesitancy expressed with the use of CLSI H62 owing to its length, broadness, cost and not being supported by regulatory bodies. The topic leaders emphasized that the document was developed by a committee made up of 37 representatives of regulators and laboratories and pharmaceutical industry and taken the effort to address over 900 comments submitted during the open review period. They also clarified it provides minimum requirements for method validation covering range of existing guidelines, and that it may ultimately provide credibility to flow cytometry as a reliable tool for high-risk endpoints.

The most prominent approach taken for validating an assay with dual use was fit-for-purpose validation (66.67%), followed by highest risk (23.81%), with only 9.52% using CLSI H62. The preferred approach to address health authority requests to utilize ‘intended-use specimens’ for validation, was to use laboratory-created specimens by 61.90% respondents, while the remaining split equally between commercially procured and in-trial (clinical verification) samples. During discussions some stated they don't use patient material but healthy donor/cell-spiked specimens for validation as patient material is not reflective of validation standards. There was agreement on the importance of working with sponsors sharing methods of practice, and including details in agreements to ensure BLA submission is not compromised. Emerging challenges highlighted included the expectations from regulators and the move towards not accepting laboratory-created specimens, which require plans to be made upfront. The general recommendation noted, while procurement difficulties were acknowledged, was to utilize a mix of native and laboratory-created specimens based on regulatory interactions.

The next portion of questions focused on LLOQ, with the preferred method of choice to create a sample for LLOQ voted as to deplete the target population (by 71.43% of respondents), while the most popular approach for establishing LLOQ for an assay with multiple reportables was LOQs for lineage-specific markers but not their derivatives, with over three-quarters of the votes (76.19%). Spectral flow cytometry is a relatively new technology with LLOQ being a streamlining process leaving what is valid. Comments from discussion included some stating they do not receive many requests for lineage-specific markers, while others highlighted that biomarker-level validation was easier, but not applicable to flow cytometry. It was agreed there would be more challenges to be faced by all.

Titration of target cell type in suitable matrix was the most voted best practice (85.71%) for demonstrating detectability 1-log below LOQ to address US FDA recommendations for MRD assays. Of the respondents, 71.43% stated they derived LLOQ for cell frequency and 76.19% voted the most used acceptance criteria for LLOQ as being ≤35% CV.

Further questions on validation approaches pursued with minimal discord. A total of 85.71% validated their assays for minimum acquisition counts. When asked about the approach followed for new lot of antibody cocktail and critical reagent validation, 61.9% used ≤0.5 log difference from the original result, while 33.33% used ≤25% CV from original result, both established during validation. When demonstrating inter-instrument precision in a clinical laboratory with standardized instruments, the preferred approach by the majority (71.43%) was to utilize two instruments and extrapolate the results to others, while 19.05% validated on all available instruments dedicated to the trial. Concerning stability, the most popular approach employed to demonstrate specimen stability was contrived or disease specimen (71.43%), and the vast majority (85.71%) performed short-term antibody staining stability for batched specimens acquired on same day of staining.

There were variations in approaches when exploring the methods of choice for demonstrating analytical accuracy. While the majority used orthogonal methods/technology (66.67%), 19.05% stated they did not perform these in their laboratories or that they used commercially available reference materials. It is noteworthy that no one used diagnostic grade assays with similar properties. With surge in use of AI in the industry, it was no surprise that a total of 80.95% respondents used AI-based analyzes for reporting clinical trial results (76.19% for manual gating and 4.76% for clinical data). There was a clearly preferred negative control, namely, 80.95% utilizing FMO(x). With the same percentage, the preferred positive assay control during acquisition was healthy donors. There was 85.71% agreement all results had to go through QC for an MFC assay with over 100 reportable results, and 95.24% use a dedicated protocol for performance qualification of a new (spectral) instrument.

Finally, an overwhelming 85.71% indicated interest in directly contributing to the harmonization of flow cytometry assays validation among GCC CROs with the session closing with discussions on how to work together to achieve this at the upcoming GCC meeting focused on flow cytometry with the session leaders agreed to lead the upcoming GCC meetings.

3.5. Biomarker assays for tissues: matrix, long-term stability & assay validation

While most CRO laboratories are set up to support regulated studies, playing a crucial role in validation of various types, there are limited guidelines regarding biomarkers, leaving the establishment of a solid framework to individual laboratories and scientists. This lack of specific guidance raises concerns, as these laboratories recommend suitable approaches to support critical decisions, often with minimal access to prior knowledge about the properties of the drugs. Twenty representatives from GCC companies participated in a 15-question survey on biomarker assays for tissues, focusing on matrix, long-term matrix stability and assay validation.

Of the 20 responses, equal numbers of respondents (75%) reported experience using chromatographic assays and ligand-binding assays (LBA) for PK, while 20% reported experience with LBA for antidrug antibody assays (ADA) for analysis of tissue matrices. Cynomolgus monkey tissue were the most used matrices for biomarker assays (85% of the respondents), followed by human (75%), rat (70%) and canine tissues (25%).

Clinical protocols and the assay purpose often drive decisions about tissue sample types to be used. Freshly frozen (FF) samples were used by all respondents, while only 15% indicated using formalin fixed paraffin-embedded (FFPE) tissues. This disparity was attributed to the challenges of using FFPE in bioanalysis (e.g., interference issues). Although, it was noted that FFPE samples are relatively easy to use for small molecule bioanalysis and that more reasonable workflows (recently developed) can reduce the challenges associated with these samples.

Mechanical homogenization was employed by half of the respondents for tissue homogenization, with 45% using a combination of this, alongside sonication and/or digestion. Considerations when determining the technique employed were discussed, such as the type of tissues requiring different strengths of homogenization, whether extensive protein breakdown was required, and the use of additional chemicals. Almost all respondents used tissue matrices for fit-for-purpose validation (95%), with just under half using them for method performance evaluation (without QA; 45%) and full validation (with QA; 40%). Confusion and discrepancies surrounding these terminologies, especially with variations from a biomarker perspective, were noted. A broad range of biomarker molecules are being measured from tissue matrices: proteins and peptides (80%), small molecules (70%, e.g., amino acids, hormones, steroids, etc.), large molecules (60%, e.g., antibodies), oligonucleotides/nucleic acids (35%), cell and gene therapies (30%, e.g., enzyme replacement, CAR-T).

The next set of questions delved deeper into the technicalities and challenges of using tissue matrices. For assays that require the use of rare tissue matrices, 70% reported assay development with an alternate surrogate matrix, while 20% mentioned substituting the matrix with a similar tissue matrix along with additional equivalency testing. When asked, “What types of WC samples do you include in tissue bioanalysis?”, 70% of respondents reported using a combination of QCs prepared by spiking analytes in tissue homogenates and QCs prepared by spiking analytes in surrogate matrix/surrogate homogenates. However, follow-up discussions indicated apprehensions about mixing, highlighting potential concerns in this approach.

Looking specifically at how respondents evaluated long-term stability (LTS) for tissue matrices, the majority (60%) indicated that they spiked analyte in the tissue homogenate prior to assessing the LTS in frozen tissue homogenate, 25% reported assessing LTS in frozen intact tissue and homogenate. Notably, no respondents mentioned spiking analyte onto the surface of the blank tissue and assess the LTS in frozen intact tissue. Follow-up discussions emphasized that this would be dependent on various factors such as the type of assays (PK or biomarker) and how frequently LTS assessment is required for the project. If intact tissues cannot be homogeneously spiked, stability assessments are often performed in tissue homogenate. Ideally, the homogenization and storage of tissue samples from dosed animals should occur as soon as possible after collection as indicated by 90% of respondents who reported performing tissue homogenization at the bioanalytical laboratory, rather than at the clinical site (0%). When homogenization of tissue samples cannot be done immediately after collection, the preferred storage temperature for intact tissue was overwhelmingly voted as -70°C (95%).

The final section of the survey explored the guidelines and validation criteria employed. Half of the respondents stated that they were using the US FDA BMV guidance for industry (2018) [22], while 35% reported using the ICH M10 (Section 7.1. on Endogenous Analytes, although it does not cover bioanalytical methods for biomarker) [21]. The remaining respondents stated that they used both FDA and ICH M10 (5%), a lack of coverage in the ICH M10 of biomarkers (5%), or that they could not claim full compliance to the guidelines owing to the partial coverage in the US FDA BMV guidance (5%). Discussions emphasized the importance of sound scientific judgement in determining appropriate validation steps and ensuring the criteria used are fit-for-purpose. There was consensus on the significance of receiving direction from sponsors and engaging in discussions about the study's needs, including whether to adopt method qualification and/or method validation approaches.

Numerous challenges related to biomarker method validation supporting tissue biomarker studies were noted, selectivity, matrix effects and parallelism were the most challenging aspects. Selectivity is particularly difficult to evaluate on an endogenous level. The majority of respondents (80%) were not conducting ISR for sample analysis using biomarker tissue assays. Variability in the acceptance criteria used for endogenous biomarker assays was evident: 15% reported using the same acceptance criteria as endogenous molecule PK assays; 45% reported using different acceptance criteria; and the remaining 40% stating they use different criteria for small and large molecule assays. While it seems logical that different studies may need different criteria, there was agreement that any future guidance providing direction on appropriate acceptance criteria for different studies would be welcome.

3.6. Evaluation of internal standard response

Internal standards (IS) are commonly used in chromatographic analytical methods to correct for variability in sample processing and analysis. Theoretically, the responses of IS from all samples in an analytical run are anticipated to be similar to each other, but, in some cases, variability in IS responses is observed and could potentially impact the accuracy of analyte concentration measurements. Laboratories often rely on a pre-set acceptance criteria specified in standard operating procedures (SOPs) to identify study samples with unacceptable IS responses. However, this may not always detect potential problems or provide adequate information about the causes of the variability. In 2019, the US FDA published a guidance titled ‘Evaluation of internal standard responses during chromatographic bioanalysis: questions and answers’ [31], which explained that when observed, IS response variability may impact the accuracy of data and potentially warrant further investigation. This session focused on the critical evaluation of the IS response during sample analysis and current industry practices, based on a survey of eight questions which was sent to GCC membership and responded by 24 CRO representatives.

The first set of questions examined the acceptance criteria used to evaluate IS responses. Almost three-times as many respondents stated that their IS response acceptance window was based on the mean IS response of CALS/QCs compared with those who based it on mean IS responses of unknown samples and CALS/QCs (62.5% vs 20.83%), while a small portion stated they set their criteria based on a statistical approach (4.17%). Some of the other methods detailed in the responses included basing IS response on an additional assessment made by comparing the unknown to known standards and calculating the mean of the unknown (provided there is cross-over of the mean of the unknown with the mean of the CALS/QCs). The majority of respondents were using SOPs for their pre-defined criteria (83.33%) while a handful were using study-specific protocols (12.5%).

A third (33.33%) of respondents stated that they would update the IS acceptance criteria mid-study if consistent re-assay values agreed with the original (±20% difference), while 37.5% would do so if unknown samples consistently fell within a certain range of CALS/QCs. Of the remaining respondents, some provided further reasons for change such as if CALS/QCs fell outside the preset criteria range, while the others stated they would not update the criteria mid-way or that it would be very unusual and likely addressed during analysis.

The next set of questions focused on how members dealt with aberrant, or variable IS responses. When exploring what approach is used for re-assaying samples with aberrant IS responses, half of the respondents re-assay the sample neat, then compare the results to the original concentration. A fifth stated they re-assayed with dilution using the matrix used for CALS/QCs for dilution, while 8.33% re-assayed in duplicate. It was discussed that approaches taken would also depend on additional factors such as if re-assaying is warranted and scientifically justified on initial assessment, and if variations were in a few, or many, samples (identifying any misalignment with method). When examining what triggered investigations into failing IS trends through a multiple-choice question, the most prominent causes were trending in several batches for the same method (75%); trending within a single subject (70.83%); trending within a batch of samples (66.67%); significantly variable individual responses (62.50%); and mismatching values when samples are re-assayed for failed IS variability (50%). The importance of the specific circumstances and good scientific judgment in triggering investigations, and how this should ideally be updated in SOPs was highlighted in follow up discussions.

How the respondents react when variable IS responses are observed in matrix effects evaluations during validation was explored in the next question, where a third responded that they accepted the experiment if accuracy was acceptable (33.33%), and 29.17% went on to investigate the variability even if accuracy was acceptable, with 16.67% stating that variability would trigger changes to the assay. The remaining 20% provided specific responses including, again, the importance of using sound scientific judgement on which option was best to take in each specific situation, as well the importance of the stage at which matrix effects should be validated (method development or method transfer). Further discussions highlighted that observation of IS variability might trigger different acceptance criteria for IS (which would be described in the study plan), and that the causes must also be considered, such as whether the method was robust enough or whether the matrix effects could be the reason for IS variability. If changes to the method are warranted, they would ideally be investigated at method development stage.

In the concluding question, almost half of the respondents (47.83%) stated that they had examples of variable IS responses impacting sample concentrations with an analogue, with roughly a fifth (21.74%) having examples with both analogue and stable label IS. When discussing how the industry was setting the acceptance window and whether they differed for analogue IS and stable IS, it was concluded while the most standard used was 50–150%, the window vastly varied between laboratories and CROs. It was agreed that the definition of the window in the study and SOP was critical.

3.7. Current disagreements in immunogenicity harmonization

As immunogenicity assay performance expectations continue to evolve, so does the need for harmonized bioanalytical ADA validation testing and reporting. Recent harmonization efforts with cross industry contributions from industry and regulators have resulted in publication entitled White Papers on ‘Anti-drug Antibody Validation Testing and Reporting Harmonization’ [32], ‘Anti-drug Antibody Sample Testing and Reporting Harmonization’ [33], and Neutralizing Antibody Validation Testing and Reporting Harmonization [34]. While these manuscripts provide consensus recommendations related to immunogenicity validation and sample testing and reporting, disparities in practice persist, which were explored through a 20-question survey answered by 17 companies' representatives.

There can be hesitation from sponsors in adopting new harmonized recommendations as each organization may have its own internal SOPs approach immunogenicity testing and, as such, require justification for adoption of new guidance. This was evident with less than 10% stating they plan to adopt fully the method and validation protocols outlined in the 2021 AAPS White paper and just over 65% stating they will adopt it to some extent, but notably roughly 25% indicated they do not currently plan to adopt. This was also apparent when exploring what drives the number of unique individuals included in cut point datasets specified in each validation plan, with equal respondents basing it on internal SOPs and regulatory guidance (35% each), and the remaining on sponsor instructions (23%). Discussions emphasized that sponsors generally drive the design including number of biological samples required, which was generally never increased (almost 60%), while simultaneously decreasing the number of repeated measures from each sample with details made clear in validation plan. Exceptions were shown to be made for special occasions (e.g., rare matrices or vaccine studies).

In situations with high prevalence of pre-existing antibodies (e.g., anti-AAV or anti-PEG), the most popular approach taken by over half of the respondents (∼55%) was obtaining cut-point data to pre-screen individuals then proceed to analyze only less reactive (presumed negative) samples in the validation cut point exercise, while 30% voted they analyze samples without pre-screening individuals to best capture the prevalence and spectrum of pre-existing reactivity within the population. For distinguishing individuals with pre-existing antibodies from individuals presumed to be negative for pre-existing reactivity in such situation, roughly 60% voted they leverage arbitrary thresholds (e.g., normalized signal greater than ∼2.0 or %Inhibition greater than ∼30–50%), while 23% fitted a bi-modal (or other multi-modal) distribution to the population to guide identification of a putative positive/negative threshold. Other approaches were discussed (e.g. partition method).

For in-study cut points there was variation in typical approaches, with a roughly similar split using available sample size and/or other factors and conducting repeated measures for baseline/pre-dose samples to best mirror the balanced design used during validation, and just over 20% measuring each in-study baseline/pre-dose sample only once to obtain a reportable result and to assess need for and/or calculate an in-study cut point. Discussions revealed multiple factors determined the approach and calculations used, such as sponsor need and budget, and validation requirements.

Even when a development program is not expected to enrol normal/healthy individuals, over three quarters of the respondents establish a cut point using normal/healthy individuals in addition to any relevant disease-state(s). For oncology basket trials (often sponsor directed), over 70% stated their approach to disease-state cut point(s) was establishing it only if individuals from a specific tumour sub-type are determined to be significantly different, while the remaining voted that each tumour sub-type should be considered a unique population and therefore always require specific disease-state cut points.

Singular LPC concentration for common use in both the screening and confirmatory assay tiers were routinely or by default established by majority of respondents (58.82%), and if sensitivity of the screening and confirmatory tiers within two-fold of each other was established by just under a quarter of the respondents, again often as directed by the sponsors. There was variation in the general approach taken for establishment of system suitability criteria for ADA/NAb assays, as they are SOP- and circumstance-dependent. While a roughly equal split of respondents used upper acceptable limit on raw signal of NC and lower acceptable limit on normalized signal of LPC (37.50%), and dose-dependency of PCs (31.25%), upper and lower acceptable limits on NC and all PCs was indicated by 12.50%, while 18.785% stated other approaches. Approximately 90% of the respondents stated they have not yet reported signal-to-noise (S/N) as an alternative to titer, with half of those currently discussing how to implement. Long-term stability data for ADA was not evaluated by over 80% of the respondents. When asked what was implemented if selectivity fails at LPC level, re-evaluation at a higher LPC level was the most popular response.

There was unanimous agreement on the routine evaluation of the impact from hemolysis and lipemia, with the most popular default design being the evaluation in pooled matrix at a single, representative level of hemolysis and/or lipemia (∼50%), then roughly a third evaluating in X number of individuals at a single, representative level of hemolysis and/or lipemia per lot. 100 ng/ml was the level of PC typically used to achieve adequate drug tolerance.

ADA assay comparability assessments across laboratories were noted to be driven by sponsors by half of the respondents and by internal SOPs by a quarter, with just under 20% stating they had not conducted these. Just over 20% in total analyzed in-study samples at each site in addition to surrogate PC and/or NC (at 25–50% or strived to analyze them) during comparability assessments, and half of the respondents were not analysing them and just using surrogate PC and/or NC. It was noteworthy than over 30% had not conducted cross-site comparability assessments and discussions revealed there could be global disparities on the approaches taken.

3.8. Changes to the bioanalytical industry: staffing, supply chain, capacity/lead time & costs

The COVID-19 pandemic has left lasting changes on how businesses and companies are run across all industries globally, and the bioanalytical industry is no exception. As the landscape continues to shift, this session explored the impact of the pandemic on bioanalytical operations in CRO settings. A survey consisting of 21 questions was sent to GCC members with 25 companies' representatives responding.

Remote working has become a standard option for many since the pandemic, although trends continue to change. Comparing the percentage of staff working remotely now versus pre-pandemic levels, just over half of the respondents (52%) stated that 5–25% more staff worked remotely while just over a quarter (28%) stated the levels were about the same. There was general agreement that remote working was role-dependent with laboratories evaluating the needs and requirements of each role. Largely, laboratory roles are required to be onsite, while supporting staff have variable modes of work. This was reflected in the results of the next question, where the frequency of bioanalytical principal investigators (PIs) supporting GLP work being onsite full-time was found to be 58.33%, and those in 3–4 days a week was 29.17%, indicating a predominantly onsite mode of working. It was agreed that there is more flexibility compared with pre-pandemic levels with amicable agreements made between teams, as the PI role often requires running of sites and responding to issues in real-time requiring presence. The benefit of the presence of leadership and experience onsite to support new talent was also cited as a key reason. No efficiency issues had been observed with the electronic and digital evolution of modern laboratories, although caution was raised about the cumulative effects of the lack of personal interactions and subsequent learnings.

The impact of the pandemic on recruitment of laboratory personnel was investigated in the next set of questions. For laboratory, and support staff, the quality of resumes submitted post-pandemic remained largely the same (52 and 60%, respectively), with just over a quarter noting slightly weaker submissions (36 and 28%, respectively), although improvements in resumes were noted for neither role. For PIs however 4% stated the resumes had improved, with 60% stating the quality remained about the same as pre-pandemic levels, with a total of 36% indicating weaker resumes were seen. The discussions raised concerns regarding the reasons for weaker resumes seen in higher-level roles including shorter tenures, high drop-out rates and low levels of in-depth knowledge. The impact of the pandemic itself was considered as a cause, where in the UK for example, applications from students having just completed studying remotely during the pandemic, had dipped in quality and presented skillsets owing to the lack of opportunity to build them in social settings.

Changes in recruitment timelines since the pandemic for the different roles were explored for which the response patterns varied for laboratory staff, support staff and PIs. For laboratory staff, significantly different views were indicated with a variable split for no (32%), 1-month (16%), 2-month (28%) and >3-month delays (24%). Discussions indicated it was not unusual to see delays of 2–3 months even during pre-pandemic times in some instances due to economic changes in the biotechnology space, which had evolved into pandemic effects and, now again the economic struggles seen globally. On the other hand, those with no delays detailed reasons such as investments, use of efficient HR teams and geographical benefits (such as, for example, the location of Australia enabling it to retain its continuous flow of university graduates). The picture was very different for support staff (predominantly involved in reporting, QC, sample management etc.) where not a huge impact on recruitment timelines was indicated (44% indicating no and 28% indicated 1-month delays). For PIs the results were somewhat surprising with 48% indicating no delays while 28% indicated >3-month delays. Long lead times have historically been required for recruiting PIs – somewhat diminishing the observed impact on recruiting for these roles. The discussions attributed some of the extreme variations to differences in company sizes and resources, also highlighting the impact of international recruitment, which inevitably will have longer lead times. It was also noted that there is a possibility that discrepancies in lead-times may not in fact be attributed to pandemic effects.

A significant and real change however was observed in salary differences in the marketplace post-pandemic. With a staggering sum of 88% of the respondents stating they saw between a 10 and 50% increase in salaries since the pandemic, the global demand for applicants is evident. With knock-on effects on prices, supply, and profit margins, this could potentially change the industry's landscape. Discussions indicated yearly adjustments to the financial markets were reflected in the results, with many companies practicing regular review of their recruitment and HR strategies, as well the evolving needs and demands of employers to balance retention and recruitment. Exploring the changes in working patterns with sponsors, cancellations or holding of programs was observed to occur at about the same (32%), slightly more (32%) or significantly more (28%) levels in general. It was clear from the discussions that this pattern had a detrimental impact on finances with investments made in developing and validating assays. With staff built specially around programs, the impact on training and practice was also highlighted. In discussing how the industry can adjust, there was general agreement that utilization of cancellation fees should be included upfront in contracts to compensate for the cumulative impact felt because of cancellations/holdings of programs. In terms of sponsor delays since the pandemic, for example with materials, availability, signature on directives and the likes, just over half of respondents stated there were slightly more delays (56%) with a one-fifth stating they saw significantly more, and an equal split stating the levels were about the same.

Looking at laboratory practice, while current capacity compared with pre-pandemic levels were indicated to increase by a sum of 56% of the respondents, 28% stated capacity remained around the same. The impact on lead times appeared to vary significantly, with 44% indicating the lead times remained about the same since pre-pandemic levels, yet 20% stated a 20% increase, and 24% stated a 50% increase. Despite the surprise among participants in this variation, there was a general consensus that capacity had been reached, and that lead times will likely vary between, for example, small-molecule and large molecule work. The next set of questions explored how often delays were encountered from historical timelines, which would significantly impact businesses. For receipt of common matrices (e.g., EDTA, rat, mouse, dog or human), the majority saw delays occasionally (37.50%), or about 20–50% of the time (41.67%), while for special matrices (e.g., specialized populations, screened matrix or non-standard species), the majority of saw delays between 25 and 50% (52%) or <50% (28%). These were noted to cause substantial problems, impacting timelines and schedules. For historical timelines for receipt of chromatography supplies, extraction supplies, equipment, and equipment services, there was a general pattern observed with most indicating occasional delays, followed by 25–50% delays, then rare delays, with a small portion indicating over 50% delays. Receipt of critical reagents however, swayed a little from this pattern, with very fewer indicating rare delays than over 50% delays, which was a surprise to some, and it was noted that the delays may be dependent on the type of critical reagents.

The session closed looking at the overall impact observed on pricing of materials, with 96% of the respondents indicating increases. Detailed responses on specific items seen with significant increases (>25%) showed that matrix (especially primate; as well as other animal and rare matrices), consumables (e.g., plates, plastics, SPE cartridges and solvents) and reagents (critical, MSD and PCR reagents, and kits) were the main issues, impacting pricing and timing. The survey and discussions indicated that the pace of drug development and expectations of sponsors in terms of timing and pricing have returned to (and in many cases, beyond) levels observed before the pandemic. Simultaneously, the landscape has shifted dramatically in terms of staffing and consumable timelines and costs. The demands of adaptive clinical trials and ever-increasing regulatory requirements dictate that the industry must also adapt as the ‘old’ way of doing things is no longer sufficient.

4. Future perspective

The GCC as a global organization will continue to provide recommendations on hot topics of global interest in bioanalysis. Please contact the GCC [35] for the exact date and time of future meetings, and for all membership information.

Financial disclosure

The authors have no 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.

Competing interests disclosure

The authors have no competing interests or relevant affiliations with any organization or entity with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, stock ownership or options and expert testimony.

Writing disclosure

No writing assistance was utilized in the production of this manuscript.