Accelerated Proof-of-Concept Q&A
Execution of an accelerated proof-of-concept study presents logistical challenges including:
• Dosing as many as three parts of the study simultaneously,
• Recruiting patients for the proof-of-concept cohort of the study, while the first cohort is ongoing,
• Coordinating clinical, bioanalysis, data management, and pharmacokinetic analysis so informed decisions can be made by the safety committee.
An experienced study team is required to successfully manage these challenges, with each member working closely with one another. At Cetero, we assign a senior clinical study manager to work with the principal investigator to monitor all clinical aspects. In addition, an experienced project manager is responsible for on-time and on-budget delivery of the core study components, including clinical, bioanalysis, pharmacokinetic analysis, data management, and biostatistics. Ultimately, this oversight ensures the accelerated proof-of-concept design decreases study time and reduces costs.
Based on Cetero’s experience conducting accelerated proof-of-concept studies, the IRB approval process has been similar to standard first-in-human study approvals. As in first-in-human and proof-of-concept study designs, IRBs have established safety parameters to consider in accelerated proof-of-concept studies. In our experience, the main challenge with IRB approval involves demonstrating that adequate safety reviews are in place to transition from first-in-human to a proof-of-concept study without having to go back to the IRB. Our policy with first-in-human studies mandates a safety-committee review of data from each cohort to either ensure it is safe to proceed to a higher dose, or to proceed to the next part of the study (i.e., continuing from single dose to multiple dose). More questions often arise from the IRB with accelerated proof-of-concept designs, so allocating more time for discussion with the IRB is recommended.
We recommend the following best practices in the development and validation of biomarker assays:
- Establishing parallelism of incurred samples for quantitative assays. Parallelism is the process of diluting samples at various ranges to establish if all the values remain parallel throughout the dilutions. For an assay to be quantitative, parallelism should be demonstrated.
- Establishing sample controls for assay implementation (production phase). If feasible, sample controls, which comprise incurred samples at low, medium, and high levels of the biomarker of interest, should be established and their assigned values determined during assay validation using multiple replicates in multiple runs.
- Careful evaluation and establishment of sample collection, processing, and storage before assay validation occurs should be noted. This process will prevent the quantity of detected forms from being changed due to accidental degradation or activation of biomarkers during the sample handling process.
- Measuring the reference interval of the specific biomarker in the targeted subject population and determining the effect of therapeutic intervention. This practice allows for an accurate estimation of required assay sensitivity to be made. While lower limit of detection can be used to judge assay sensitivity for some biomarker assays, it is a good practice to use lower limit of quantification to measure assay sensitivity for biomarker assays, especially the assays intended for advanced validation.
The most important stage of biomarker validation is the determination of specific biomarkers you need and what you plan to do with the data. Only after these parameters have been established can you select the proper degree of validation for each biomarker assay in each study using the fit-for-purpose guidelines. Fit-for-purpose assay development allows you to conserve your precious resources when conducting early phase clinical studies where biomarkers are still being selected, yet will provide the rigorous degree of validation for later phase studies or for critical business decision making.
Biomarkers typically fall into one of two groups: either poorly characterized, or well characterized. Depending on the degree of validation required, if at all, the validation process can be divided into stages depending on which group the assay is associated with. Assay validation for poorly characterized biomarkers typically consists of two stages: pre-validation development and exploratory validation. Well characterized biomarkers have three stages of validation – pre-validation development, exploratory validation, and advanced validation. Each stage has distinct objectives to establish. For example, during the pre-validation method development stage, the expectations for assay sensitivity, precision, and accuracy will be determined and a feasibility procedure for sample collection, processing, and storage will be established. As the assay moves into the exploratory or advanced assay validation phase, data are collected to demonstrate that the assay performance meets the requirements. These stages constitute an integral and evolving process of biomarker assay validation and are all dependent on the initial decision about the statements, or actions based on the data will be made.
Because population PK/PD modeling is now being used virtually in every submission, specific therapeutic areas won’t necessarily benefit more from this approach. Instead, it may be simply easier to perform population PK/PD modeling within specific therapeutic areas.
In some instances, when the correlation between PK and efficacy and toxicity are very well known, such as in infectious diseases, PK/PD modeling is easier to perform. On the other hand, the mechanism of action of many drugs against neurological diseases such as Alzheimer’s disease may not be fully known. Biomarkers describing the potential efficacy are therefore more difficult to find. The true clinical end point may also take years to achieve. All these factors will add challenges to population PK/PD modeling.
When considering placebo response and drop-out within PK/PD analysis, it is considered a best practice to model exactly what has occurred and what is happening within PK/PD relationships. The ideal PK/PD model will have a placebo response built into it. If a participant drops from the study, then it is advisable to model exactly what has happened. Regarding the modeling of the placebo response, it is essential to have a good understanding of the mechanism of this response. An example of this involves working with a nasal delivery drug. If you administer a nasal placebo containing a saline solution, the saline will wash out the allergens in the nasal passage, thus causing a true physical response in addition to the placebo response. Understanding this mechanistic basis is essential in creating a PK/PD model that takes the placebo response into account.
The major key criteria behind the appropriate level of validation for a biomarker assay is the knowledge of whether or not this biomarker will become “pivotal” from a regulatory point of view in the pharmacokinetic (PK), safety, or efficacy section of the submission. In other words, if the biomarker itself becomes the efficacy measure for the Phase III trial, then it needs to be fully GLP validated. If the biomarker is only used as a marker of efficacy or safety to better understand the optimal dosing regimen, then it does not need to be GLP validated. The FDA and EMEA are now meeting jointly to discuss the clinical validation of biomarkers. However, the need to use a biomarker that the FDA has deemed to be clinically validated is only needed if the biomarker will become pivotal. POC studies often include analysis of biomarkers that are used for research purposes, so their measure does not require a bioanalytical validation or for the FDA to consider them clinically validated.
Developing accelerated proof-of-concept (AcPOC) protocols can be very challenging, as they require all the clinical pharmacology and medical experience and expertise that go into designing, conducting and analyzing data from single ascending dose (SAD), multiple ascending dose (MAD), and proof-of-concept (POC) studies. However, AcPOC designs are made even more complicated by the fact that these studies are run concurrently. Early cohorts of the MAD study may be starting while late-stage cohorts of the SAD study are still being conducted. Therefore, the protocols need to have the flexibility to allow the designs to be adapted as each stage progresses. As with many early clinical development programs, protocol amendments may be needed, since the pharmacokinetic parameters used in the design are not based on human data but extrapolated from the preclinical data. This is why Cetero often makes our protocols more flexible. From the beginning, they are “adaptable” in their design and in the dosing regimens to be studied.