Rewriting the rules for untargeted biomarker discovery

Dr Jonathan Lawson at Owlstone Medical presents key lessons related to priorities, quality management and study design that have been drivers in developing chemical detection technologies.

The potential for biomarkers in early detection, drug development and precision medicine is undeniable, and it is no surprise that this has generated a rapidly-growing market in excess of $50 billion with more than a million scientific papers published, over half of them in the last decade. A large part of biomedical research is now focused on finding biomarkers across a range of sample types and analytical approaches. Unsurprisingly, increased interest in biomarkers has led many to invest in untargeted biomarker discovery studies aimed at uncovering previously unknown factors that can be reliably related back to clinically relevant biological changes, such as the onset of disease, differentiation of disease phenotypes, or responses to treatment. However, few of these have resulted in success and these studies are now often compared to hunting for a needle in a haystack, or to fruitless fishing trips – largely dependent on good fortune.

Non-invasive biomarker analysis

Breath Biopsy from Owlstone Medical provides non-invasive biomarker analysis of volatile organic compounds (VOCs) collected from breath (Figure 1) and is being developed to provide more consistent solutions for biomarker discovery. The technology has its foundation in over 15 years of experience providing reliable, high-sensitivity chemical detection solutions for industrial and military applications. Breath is an emerging biomarker matrix with considerable clinical potential due to the capacity for non-invasive sample collection, which is compatible with most healthcare contexts and could even enable at-home health monitoring.

Several key challenges that have negatively impacted the chances of success in the field have been investigated, including lack of consistent methods and reporting, the need for appropriate standards and controls, and the requirement for appropriate optimisation and quality control at all stages of sample collection and analysis. A simple study design has been developed at Owlstone Medical called Breath or Blank (BoB) for comparing methodologies, allowing for the quantification, and reporting on improvements in biomarker discovery, that are independent of the techniques used.

Biomarker study planning and reporting

Suitable planning and reporting of biomarker studies are equally as important as the quality of the study itself. Breath biomarker studies often encompass hundreds of compounds, while ‘omics studies may consider thousands. Despite this scope, many studies, particularly in breath research, are small—often pilot studies with tens of samples—and use inconsistent methods and reporting practises. Studies with many potential candidates and few samples to analyse have a much greater chance of discovering false biomarkers. A p-value of p<0.05 means a 5% false discovery rate, which could equate to hundreds of compounds and, as such, even with a very large number of samples, it can be difficult to reliably identify true biomarkers. During study design, tools such as power calculations, should be encouraged, to understand the level of confidence that can be ascribed to the results of a study of a certain scale. Additionally, using reporting structures such as the standards for reporting diagnostic accuracy (STARD) guidelines can greatly improve the comparability of results across studies. Structured reporting in turn allows comparison to more established biomarker fields, and meta-analyses across studies that effectively boost the value of small-scale studies.

Breath collection devices

Biomarker sample collection can be a challenging process, and this is particularly true for non-invasive breath sampling. Until recently, many breath sampling devices for biomarker discovery have collected breath into bags or tubes, resulting in limited and inconsistent sample sizes, as well as issues with sample contamination, transportation, and storage. With over 100 breath research leaders in the Breathe Free consortium, a breath collection device, the ReCIVA Breath Sampler, was developed that prioritises standard sample collection and facilitates reproducible results. Focusing on sample consistency, quality, storage, and the exclusion of contamination is vital to developing technologies suitable for mainstream clinical use. Combining ReCIVA with a portable air supply has allowed significant reduction in contamination from VOCs in inhaled ambient air (Figure 2). Just as the Vacutainer enabled widespread blood testing, so too can devices like ReCIVA advance breath sampling.

Biomarker analysis

Biomarker analysis tools vary greatly depending on the types of biomarkers being studied. For VOCs, the gold standard is gas chromatography mass spectrometry (GC-MS), which offers high sensitivity detection, compound separation and identification capabilities. Breath is an extremely complex biomarker matrix and over 1,000 different VOCs have been reported in human breath samples, making reliable separation essential for high confidence biomarker detection. Expanding on this capability, high resolution accurate mass (HRAM) GC-MS allows even greater separation of complex matrices, particularly when combined with a fully optimised sample processing workflow. 

Working on our biomarker pipeline has highlighted the importance of thoroughly assessing quality at all stages. Our internal system was developed to include various checks and redundancies, to maximise the chance of recovering accurate data for all samples. The process includes a barcode system that tracks each sample through every stage, allowing samples to be linked to meta-data related to sample collection, shipping, and processing. In turn, these can be used to flag potential issues to look out for in data processing.

Every set of samples also includes many standard compounds. Each sample is injected with a standardised quantity of eight deuterated internal standard compounds with diverse chemical characteristics. These standards allow us to assess the quality of compound recovery for each sample and identify issues during analysis. Similarly, every set of samples is processed alongside prepared calibration samples containing varying quantities of 52 compounds found on breath. Using these, larger-scale challenges such as instrument drift between analytical cycles can be assessed.

The future of biomarker discovery

Reliable biomarker discovery remains a challenge across the field. Addressing this requires us to step back from the eventual goal of clinical application and to carefully consider all aspects of the discovery process.

A huge diversity of discovery pipelines have been developed and this variety present barriers to biomarker development. Only by being able to compare approaches and by taking steps to understand the causes of variation, can we hope to develop biomarker discovery pipelines that can consistently produce reliable results that can be verified and progressed towards clinical application.

Breath is non-invasive making possible to apply in a wider range of contexts, including for at home use. In addition, it is essentially inexhaustible, meaning it is well suited to regular repeat testing and ongoing monitoring. The development of reliable sample collection and advances in method standardisation and reporting will greatly enhance the discovery of relevant biomarkers across diseases and applications including early detection, precision medicine and drug development.