Multiplexing Workflows for Biomarker Discovery

Biomarkers are valuable indicators of normal biological processes, disease pathogenesis, and patient responses to treatment. Yet, discovering new clinically relevant biomarkers can be like looking for a needle in a haystack, highlighting the value of a multiplexed approach. This article describes some of the challenges of finding a novel biomarker and provides suggestions for optimizing multiplexed workflows to increase the likelihood of biomarker discovery.

Challenges of biomarker discovery

According to Dr. Tobias Polifke, Co-Founder and Managing Director at CANDOR Bioscience, a fundamental challenge of biomarker discovery is that real on-off targets are rare. “It is seldom the case that a biomarker is either present or absent in the case of a disease,” he reports. “Instead, a biomarker may demonstrate an increase or decrease in concentration, or its concentration may vary in relation to another biomarker. Identifying biomarkers in a screening setting requires an assay with sufficient accuracy and precision to reliably detect such changes while at the same time addressing the inherent instability of many biomolecules in patient sample material.”

Valerie Jones, Ph.D., Marketing and Technical Support Director at RayBiotech Life, notes that a further challenge of biomarker discovery is analyzing enough samples and controls to generate statistically powered data. “All too often, potential biomarkers are identified using few samples with no follow-up study employing an independent sample cohort,” she says. “In these situations, overfitting the data is a major concern.” Discovering novel biomarkers may additionally be complicated by the available sample volume, which can limit the types and number of assays that can be performed, and by the vast range of protein concentrations present in sample material, which may span over 10 orders of magnitude.

Accessibility is a key consideration

Biomarker accessibility is critically important, both during discovery and beyond. Jason Kinchen, Ph.D., Lead Biomarker Scientist at Owlstone Medical, explains that while the most relevant biomarkers are often the most direct, invasive sample collection methods, such as tissue biopsy, are usually far from ideal. “In a life-threatening situation, tissue biopsy may be acceptable,” he says. “However, this is unlikely to hold true for early detection, where there is often a need to screen patients who are asymptomatic. Using low-risk sample matrices like blood, urine, or breath not only promises to speed up biomarker discovery, but also translates more readily to a clinical setting.”

Not all diseases are equal

In general, common and well-understood conditions like type 2 diabetes and heart disease have the most associated biomarkers, while rarer conditions such as various autoimmune and neurological disorders lag behind. One reason for this is that resources are often distributed where they can have the most impact—both in terms of patient numbers and return on financial investment. Large patient populations can also help reveal distinct disease sub-types. “Focusing discovery efforts on conditions that currently lack validated biomarkers, as well as on identifying biomarkers that reflect the full biological spectrum of a disease, is essential to develop relevant targeted treatments,” says Kinchen.

Multiplexing increases the likelihood of biomarker discovery

“Historically, multiplex technologies have always opened the door to the discovery of new biomarkers,” says Stephen Angeloni, Ph.D., Senior Field Applications Scientist at Luminex Corporation. “For genomic applications, nucleic acid arrays have enabled identification of hundreds of pharmacogenomic biomarkers—for example, by comparing patterns of gene expression in conditions of health versus disease. To date, other multiplex biomarker applications have focused mainly on protein biomarkers such as cytokines, chemokines, or peptide hormones. A number of these have been developed on the Luminex platform due to the simplicity of switching different targets in and out of a panel. Now, with the availability of systems for multiplexed genomic and proteomic analysis, it has become much easier to custom design multiplex assays for new biomarker discovery.”

A main advantage of multiplexing is that it improves the chances of identifying a biomarker by allowing researchers to cast the widest net possible. “Oftentimes, a biomarker signature of multiple proteins can result in higher diagnostic accuracy than using a single biomarker,” reports Jones. “For example, several of the few FDA-qualified biomarkers are actually a combination of proteins.” A biomarker discovery study published last year used RayBiotech’s antibody array technology to analyze 640 proteins in 630 individuals with and without tuberculosis; this showed an eight-protein signature to have the highest diagnostic accuracy, with an independent test cohort demonstrating 83% specificity and 76% sensitivity during validation.

Optimizing workflows 

Optimizing a multiplexing workflow for biomarker discovery relies on several best practices. “Pre-analytics is crucial to generate meaningful results” says Polifke. “This means optimizing the protocol used for sample collection, performing meticulous characterization of healthy and disease state donors, and paying close attention to handling conditions including temperature, transportation, and storage. It is equally important to assess biomarker stability and the potential for masking by other biomolecules. While some biomarkers require the addition of general protein stabilizers—for example, stool samples destined for cancer screening based on hemoglobin—others may necessitate the use of demasking diluents such as LowCross-Buffer to ensure accurate measurements.” Polifke also suggests sourcing validated sample material from a specialized company such as in.vent Diagnostica or Central Biohub^® to streamline characterization.

Kinchen comments that the nature of Owlstone’s technological focus has meant developing an alternative way of working. “A standard quality control uses a pooled technical replicate to identify and quantitate analytical variation,” he says. “However, because our technology involves detecting volatile compounds in a gaseous matrix—breath—generating a true pooled replicate is challenging. We needed to re-focus and develop metrics best-suited to the matrix at hand, based around evaluating signal over background, sparseness, and median variability for compounds.” Critically, while the nature of the QC sample may vary, quality metrics should always be evaluated for a multiplex assay throughout the entirety of its development and application.

Jones recommends seeking advice from a biostatistician before starting the study to determine the number of samples and controls that should be employed to obtain statistically powered data. “Biostatisticians can also help with data analysis and can suggest appropriate sample and controls numbers for the next step in biomarker research—biomarker validation,” she reports. “I would also recommend running a small pilot experiment first, with a few representative samples, before launching into a full screen. This can help optimize sample dilution and determine if the platform is appropriate based on detection range and sensitivity for the sample type.” Lastly, Angeloni comments that whatever the application, there is no need to be concerned if the multiplex platform has targets in it that you think are not needed. “Over the years, a number of my clients have raised this concern, but upon running a full multiplex panel for different applications many have found a new marker or profile they did not think would be important.”