The topics discussed during the session included cross-validation, parallelism, comparative bioavailability studies, combination drug stability, endogenous analyte bioanalysis, and dilution QCs.

Cross Validation

Bioanalytical data is foundational information collected throughout the drug development process, from discovery to approval and post-marketing. This information is used to characterize the exposure of drug candidates and analytes of interest (e.g., metabolites) relative to their safety and efficacy. Appropriately validated methods are employed in the generation of bioanalytical data. Due to the importance of the data, bioanalytical analysis supporting toxicology studies is conducted under the scope of the Good Laboratory Practice regulations and clinical studies are required to be validated per regulatory guidelines (FDA, European Medicines Agency [EMA], etc.) (2, 3), and most recently by ICH M10 (1).

The drug development process typically spans several years. Over this time, bioanalytical methods may evolve or change commensurate with the program's needs. Additionally, it is common to require methods to be implemented in multiple laboratories to support a single study. Alternatively, different methods may be used to support various studies, and the data from these studies may need to be aggregated and used for pharmacokinetic (PK) characterization. It is, therefore, critical to understand if there are differences in the bioanalytical data generated at different laboratories or with other methods to ensure appropriate decisions can be made regarding dosing regimens for the intended patient population.

Traditionally, bioanalytical scientists have relied on partial and/or full validations of different bioanalytical methods in the same laboratory or at different laboratories to demonstrate confidence in the data quality. However, such assessments are specific to a given method/lab.

Cross-validation is the process used to compare data from different methods or laboratories. Cross-validation is performed by analyzing the same set of samples at different laboratories or using different techniques. Although the various regional guidelines recommended cross-validation, they did not prescribe acceptance criteria or describe methods to assess data comparability. Subsequently, there were efforts to address this topic within the bioanalytical community and the result was a recommendation to use what essentially is considered the incurred sample reproducibility (ISR) approach and the associated ISR acceptance criteria (4).

Rather than taking a pass/fail cross-validation approach, ICH M10 has emphasized determining whether there is bias between data sets generated at different labs or with different methods, removing the need for acceptance criteria in cross-validations. ICH M10 recommends utilizing statistical approaches in determining the bias by measuring the same set of quality controls (QCs) spanning the calibration curve in replicates and incurred study samples, if available, with different methods or at different laboratories. ICH M10 recommends a minimum of 30 cross-validation samples to make this assessment and suggests the use of statistical approaches such as Bland–Altman Plots and Deming Regression to assess bias. However, it is noted that other statistical methods to determine bias may also be employed.

During the session, it was discussed whether samples used for cross-validation needed to include incurred samples (i.e., post-dose) in addition to the spiked control matrix. It was noted that the guideline indicated that incurred samples only needed to be included if available; thus, including such samples was not felt to be mandatory.

The lack of specific acceptance criteria for cross-validation has been challenging to the bioanalytical community. While having pass/fail criteria simplifies assessment, such an approach also limits the utilization of data from different labs/methods when the results fail. ICH M10 takes a broader approach to this topic by providing a path for data utilization if a bias is observed between data sets. An example of such an approach was presented during the session whereby a correction factor was used to aggregate data sets for alectinib and its metabolite when a clear bias between data generated at different labs was demonstrated statistically (5). However, values reported by the bioanalytical laboratory should reflect those generated directly by the assay used to analyze the samples; the application of such correction factors should be done by the individual who is aggregating the data from multiple labs.

The ICH M10 training material also indicates that cross-validation of methods with non-overlapping analytical ranges should be avoided within a study (6). However, samples and QCs can be diluted to fall in the low range if necessary.

In addition to specifying the need for cross-validation when different laboratories/methods are used to support the same study or when different methods using the same analytical platform are used to support a development program, ICH M10 also calls out the need for cross-validation when two different bioanalytical approaches, for example, ligand binding and liquid chromatography-tandem mass spectrometry (LC–MS/MS), are utilized during a drug development program. A similar approach concerning bias assessment can be used in such a case, and a correction factor can be applied if a bias is identified and defined between methodologies.

ICH M10 does not, however, require cross-validation in cases where new or alternative technology is the sole bioanalytical approach implemented from the onset of a drug's development.

Parallelism

Parallelism is the process of analyzing the relationship between the calibration curve and serially diluted study samples to detect any influence of dilution on analyte measurement. Further, it is a performance characteristic that can detect potential matrix effects. It is important to recognize that parallelism generally cannot be assessed during pre-study validation, as it requires samples from study subjects. ICH M10 recommends the parallelism experiment for full validation on a case-by-case basis when interference from endogenous binding proteins is suspected or when a lack of parallelism is suspected in different patient populations.

Historically, parallelism assessments for compounds with an endogenous counterpart have been employed with ligand-binding methods. However, ICH M10 does not distinguish between bioanalytical methodologies regarding parallelism assessment. During the session, there was a discussion about whether parallelism assessments were restricted to ligand-binding methods or needed to be employed for chromatography-based methods. A case-by-case basis was recommended for such methods.

On the other hand, a parallelism assessment for methods for analytes that have an endogenous counterpart is needed, regardless of the methodology employed, as such investigations can provide assurance that observed changes in response per given changes in analyte concentrations are equivalent for the surrogate and the authentic biological matrix or surrogate analyte across the whole range of the method.

Comparative Bioavailability (BA) and Bioequivalence (BE) Studies

The bioanalytical community has experienced an evolution of requirements and descriptions related to BA and BE.

The ICH M10 guideline refers to “comparative bioavailability/bioequivalence” 13 times in the text and seven times in the table associated with the guideline. Specific additional guidelines and expectations for the reporting of such studies are included within ICH M10. Although the definition of a BE study has been well established, more clarity is needed on what is considered a comparative BA study. Hence, the definition of comparative BA studies has been a point of confusion concerning the application of ICH M10.

To address the ambiguity around this topic, the use of these terms in guidelines issued before ICH M10 was reviewed.

The 2011 EMA guideline on bioanalytical method validation refers to specific requirements for BE studies, but the term BA is not mentioned (2).

The 2018 FDA bioanalytical method validation guidance refers to BA (and BE) but does not use the terms relative BA or comparative BA (3).

The manual associated with the US FDA compliance program for bioresearch monitoring, issued in 2018 (6), does refer to comparative BA and BE studies. The document cites them as follows: “New Drug Applications (NDAs) may rely on comparative BA studies or BE studies to demonstrate that a new drug formulation or a new route of administration of a drug has the same pharmacokinetic properties as a reference, marketed product” (7).

The FDA guidance on Bioavailability Studies Submitted in NDAs or INDs issued in 2022 (8) uses the term relative BA throughout the document. Although this guidance does not explicitly define what relative BA studies are, it is implied that all BA studies comparing different formulations or test conditions are classified as comparative BA studies (8).

During the development of a new drug or new treatment agent, multiple “BA” studies are conducted, especially during the early phases of clinical development. Some of these early studies may be conducted as pilot or exploratory studies for internal decision-making, for example, to characterize the absorption of drugs from different formulations being developed for potential clinical application. Such assessments may also include studying the effect of food on drug absorption of earlier formulations. The results of these exploratory studies may form the basis for the subsequent conduct of a definitive study. It is, therefore, important to identify which studies qualify as comparative BA studies and must conform with the additional requirements specified in the ICH M10 guideline.

Without clarity or proper guidance, bioanalytical laboratories, especially contract research organizations (CROs), may be overly cautious (to avoid regulatory inspection and potential citation) and execute all BA studies to meet the ICH M10 requirements, which translates to additional cost and time that may not be required. Therefore, close communication between the external bioanalytical laboratories (e.g., CROs) and sponsors (e.g., pharmaceutical and biotech companies) is critical. It is important to understand the relevance of how each type of study conducted during clinical development is used in regulatory decision-making.

The critical point to be assessed in whether or not a study is to be considered a comparative BA study under the scope of ICH M10 is to identify if data from the formulations evaluated in the study is to be considered in the assessment of the efficacy or safety of the drug under development; specifically, is the data from the comparative BA study being used to bridge clinical results from one formulation to another, thus enabling the aggregation of clinical results from studies that may have employed different dosage forms? Such studies should be treated as comparative BA studies under the scope of ICH M10. Alternatively, if the study results are intended to stand alone and be used solely for internal decision-making, for example, in identifying the "best" formulation to be used in a future clinical efficacy trial, the guidance in ICH M10 that is specific to comparative BA/BE studies do not apply.

Combination Drug Stability

Likely, the most controversial topic concerning bioanalytical method validation is co-medication/combination drug stability assessment. This topic first came to the attention of the bioanalytical community approximately 12 years ago when failure to conduct such evaluations was cited in US FDA inspectional observations. Since then, bioanalytical scientists have been uncertain when such stability assessments are necessary.

The ICH M10 guideline clarifies the matter by stipulating the conduct of stability studies for fixed-dose combinations and specifically labeled drug regimens. The definition of a fixed-dose combination is established in the US Code of Federal Regulations. It refers to the combination of two or more drugs in a single dosage form (9). Specifically labeled drug regimens, although not defined in the guideline, are taken to refer to the combination of two or more drugs administered as an approved regimen as typically seen in the areas including treatment of diabetes, hypertension, and oncology patients. Some of these approved regimens involve 4 or 5 drug combinations, which may comprise small molecule drugs, monoclonal antibodies, antibody–drug conjugates (ADCs), and other novel conjugated drugs, small interfering ribonucleic acids (siRNAs), and platinum agents (10). Treatment of diseases such as human immunodeficiency virus (HIV) (i.e., small molecule anti-viral agents) and tuberculosis (antibiotics) are also based on multi-drug combinations where specific regimens are approved for the indication.

It is not unusual during drug development that combination therapies, i.e., administration of one or more novel agents and/or one or more approved agents, are evaluated during early-phase clinical trials before a final drug regimen is determined. Generally, on a case-by-case basis that considers the chemistry of the co-administered drugs, conduct of co-medication stability is not expected during these early studies, as they do not utilize an approved regimen. Conduct of such co-medication stability studies could be reserved until the specific drug regimen for which registration will be sought is identified, hence conducting the combination stability experiment to support the approved drug regimen may not be possible until later in clinical development.

For specific drug combinations, a scientific rationale may be provided to justify not performing combination drug stability experiments, such as the combination of a large-molecule drug (e.g., monoclonal antibody) and a small-molecule drug. Additional information can be found in the ICH M10 Training Material (6).

Testing of required combination regimens should be conducted with the matrix spiked with all the combination drugs. The concentration of the combination drugs should be clinically relevant and representative of the observed/expected circulating concentrations, typically Cmax or Css (steady state). In certain situations, a lower concentration may need to be used if challenged with solubility issues when preparing multi-drug spiking solutions (10).

Dilution QC Stability Assessment

Whether dilution QCs need to be subjected to stability assessments is another topic that has received much discussion since the release of ICH M10. While the guideline does not specify including stability QCs when performing stability assessments, Section III of the guideline that deals with chromatographic methods indicates that dilution QCs should be included in runs that contain diluted samples. Furthermore, differing from past guidelines including the US FDA guidance (3), ICH M10 states that the dilution QCs should be prepared at concentrations above the range of the standard curve.

In contrast to chromatographic methods, the M10 guideline does not require dilution QCs for LBAs.

Given the need to include above-the-curve dilution QCs in chromatographic sample analysis runs, most sponsors and bioanalytical laboratories appear to have adopted the practice of including dilution QCs when performing stability experiments for chromatography-based methods, at least freeze–thaw and long-term frozen stability. This practice ensures that dilution QCs will not fail for stability reasons during sample analysis and thus ensures scientific rigor and clinical relevance, especially in situations where most of the samples are much higher than the upper limit of the quantitation (ULOQ) of the method and require dilution.

Deciding on the dilution QC concentration can be challenging, especially during nonclinical toxicokinetic studies and some early-phase human dose escalation studies. The dilution QC concentration does not necessarily need to exceed the Cmax, and in certain situations, the concentration of the dilution QC may be restricted by solubility limitations. Additional information can be found in the ICH M10 Training Material (6).

One additional point to note is that the guideline emphasizes the selection and validation of the method range to suit the expected concentrations of study samples. For example, if most of the samples are much higher than the current ULOQ, it may be beneficial, if possible, to validate a higher-range method (that could also minimize or eliminate the need for dilutions).

The Regulatory Perspective on Bioanalysis

The ICH M10 outlines the regulatory recommendations for bioanalytical method validation. However, the first question that bioanalytical scientists should ask themselves even before considering the guideline is, "Can I trust and rely on the results that are being submitted to the regulatory agencies in the bioanalytical method validation and study sample analysis reports?” If this information contains uncertainties, irrespective of whether it is recommended in the M10 guideline, then there are likely aspects that need to be addressed before submission.

Before submission, a best practice is to view the material to be submitted to a regulatory agency from the perspective of the regulatory reviewer. Consider what items/information a reviewer would be focusing on. Use this assessment to proactively identify what questions a reviewer may have for the sponsor.

Furthermore, it is critical to understand the impact of bioanalytical data and issues on regulatory decisions. Therefore, the project team must understand the significance of bioanalytical issues as they impact projects. There should be clear and close communication within the project team to avoid a disconnect between the bioanalytical scientists and those of other disciplines on the project team using/interpreting the bioanalytical data.

Historically, bioanalytical scientists develop and validate methods in response to the project team's requests. However, given the increasing complexity of drug modalities and the associated methodology used for bioanalytical methods, the bioanalytical scientist should be fully aware of questions to be addressed in regulatory decision-making and contribute to discussions dealing with the methods needed to address these matters. Many method characteristics can be assessed during method development, validation, and sample analysis, but they are not always required in regulatory guidelines. The bioanalytical scientist should consider drug development guidelines in addition to those pertaining to bioanalytical method validation to ensure that the method is robust enough for its intended purpose.

The speaker from the US FDA presented some common issues found during the regulatory review regarding endogenous analyte bioanalysis (11). In some cases, while a surrogate matrix was used to prepare calibration standards (CSs) and QCs, the endogenous concentration of the analyte was not accounted for in the study sample analysis. There were cases in which absence of a parallelism test addressing the potential matrix effect and differences in recovery between the surrogate matrix and the authentic matrix were observed. In some cases, the stability of the analyte during sample collection and handling was not adequately demonstrated during bioanalytical method development and validation. There were occasions when study samples went through a different sample preparation method compared to the calibration standards and QCs, resulting in uncertainty regarding accuracy and precision. These findings highlight the importance of analytical reliability as if you cannot rely on the data generated in the bioanalytical laboratories, interpreting the data in the bigger picture (i.e., clinical context) of drug development would be meaningless.

While guidelines can provide a general framework and direction that should be applicable in most situations, additional considerations may be warranted in some specific situations depending on the complexity and the context. Hence, submissions should openly address deviations from the general recommendations and principles. Justification for the deviations, along with supporting information and data, should be included in the submission. Sponsors are encouraged to proactively discuss with and seek feedback from the appropriate review division at regulatory agencies early in drug development. Early communication between the sponsor and the regulatory agency is among the most critical things in drug development. Recently, there has been an increase in dedicated questions regarding issues and challenges related to bioanalysis in regulatory milestone and guidance meetings (e.g., pre-IND meetings, Type C and Type D meetings) between the sponsor and the US FDA. There are many challenges, but it is important to remember that those challenges also provide us with opportunities.

Overall, it is important to remember that the PK and PD data should be reliable to allow clinical interpretation. Therefore, the bioanalytical methods are critical to provide the confidence in the measurement of concentrations for determination of PK/PD parameters. However, to have confidence in the concentration values measured, there needs to be confidence in the bioanalytical methods used to measure those concentrations. Therefore, it is not an overstatement that bioanalysis is the foundation of drug development (12).