Protein biomarkers provide a dynamic and real-time window into human biology and disease. They are increasingly used across all stages of the drug development pipeline, from early phase biomarker discovery and understanding mechanisms of action to patient stratification and label expansion.
In this SelectScience webinar, now available on demand, a panel of drug discovery experts discusses how they are using protein biomarkers, the challenges protein biomarkers have helped them overcome, and how biomarkers have contributed to accelerating drug development.
Read on for highlights from the roundtable, or watch the full webinar on demand, at a time that suits you.
Starting the discussion, Dr. Magnus Althage explained the benefits of using an unbiased approach to studying genetically driven diseases. Althage, highlighted the benefits of the multi-biomarker panels available from Olink from a functional and efficacy perspective, but also in identifying patients likely to respond to treatment and detecting safety signals in clinical trials. Dr. Katerina Pardali continued the conversation by illustrating how multiplex biomarkers can be used to better understand disease and intervention. Presenting two examples investigating the inhibition of MK2 mediated inflammation and immunomodulating T2 inflammation in asthma. Pardali shared how their mode of action hypothesis were confirmed and rejected through protein biomarker data. Continuing the discussion on integrating protein biomarkers in drug development, Dr. Michael D. Howell referenced research using CXCL10 as a pharmacodynamic marker for JAK inhibition in vitiligo. Howell also discussed research looking at the identification of unique biomarkers that predict response to Itacitanib in acute graft-versus-host-disease (aGvHD) before handing over to the panel discussion, highlights of which follow.
MH: We did this in a recent paper, we went through and identified protein biomarkers using Olink Explore in an early phase 1 trial. The first step was to compare different methodologies. We validated that we saw similar results across multiple platforms and then we worked with Olink Proteomics specifically to refine the assays, getting limits of detection for each of the analytes to then evaluate them further in a phase 3 trial. It was about a year-long process in terms of the overall validation, but at the end of it, we understood exactly what our coefficients of variants were for each one of the assays.
MA: When you validate your biomarker, it needs to be in the patient population because you don't know the impact of the biological variability otherwise. You need to evaluate the biomarker over time both between individuals and in the same individual, and this is something that we implement constantly as part of our validation.
KP: The assays you're using, and the questions you're asking will have to be fit for purpose. If you just want to see whether you're influencing your hypothesis, you might not need to have the most validated assay. If you are developing a biomarker with the ambition to use this as a diagnostic, you need to have a lot of validation both on the different technologies as well as on different patient cohorts.
KP: It will depend on what type of disease area you are working in. If you are looking at more chronic diseases, the accuracy of predicting the response to treatment or even phenotyping the patient to the disease would be the main objective. In other disease areas where time might be critical, like in critical care, the result turnaround time would be one of the main concerns. To be able to get a fully analyzed proteomic signature, which can guide your next step within half an hour, would be of paramount importance to a clinical care setting.
MH: Every disease we look at, for the most part, is multifactorial. These are complex disorders that have genetic, environmental, and all kinds of other factors that play in. As you're learning more about the mechanisms that drive the disease pathogenesis, and then adding that there are multiple approaches therapeutically, there are a lot of variables to consider. And so, by the time you get to predict the response of, for example, 100 different medications that are being evaluated for a disease, each one of them is potentially going to have its own predictive signature. This illustrates the challenge that we face in having to juggle many variables when trying to identify predictive biomarkers.
KP: If you compare the cost of doing a biomarker panel analysis for a few hundred patients and a few hundred samples, to the overall cost of a clinical trial, it is still inexpensive. And the downside of not doing a biomarker panel analysis is that if your clinical trial fails it may be more difficult to understand why it failed.
MH: The area of biomarkers over the last two years in the middle of a worldwide pandemic has become even more important. Olink Proteomics carried out a study where they were evaluating individuals that had been infected with SARS-CoV-2 and started looking at protein biomarkers that were elevated in those individuals. When you think where we might see biomarkers contributing in the future it's not necessarily just a human health question, understanding who was infected with a particularly life-threatening virus, it's thinking about how this impacts our day-to-day operation within medicine as a whole. Could we use those biomarkers to not only understand who was infected, why they were infected, and what their prognosis was with the disease, but could these also be used to help predict who's going to respond to the different therapeutics? Whether it's an antiviral or a monoclonal combination that could be used in those patients. It could be used to understand the efficacy of why certain patients had better responses to the vaccines than others. I think that this is where you're going to start seeing a broader application of these types of approaches as we try to understand diseases even further.
MA: An area I would highlight where I see these approaches being implemented is when we are identifying new targets. Many of these could be genetically identified targets whose mechanism we have limited understanding of. I see this broader assessment generating multiomics data a valuable tool to help identify the patient population and better understand the mechanisms of these genetically identified targets.
KP: The beauty of protein biomarkers is that they can help you understand the mechanisms that will differentially affect patients suffering from the same disease. In the future, I think that phenotyping and molecular phenotyping of each patient will become even more relevant in order to provide the most effective personalized care.