How omics methods could supercharge biomarker discovery

Treating cancer is complex. Not only is the pathology diverse, but there are many different cancers and many more individuals living with them. Doctors have long known that matching therapies to both the disease and individual often yields better outcomes than treating the disease alone. That is the basis for precision medicine, which examines various biomarkers to help guide therapy and prognosis.

In oncology, development of precision medicine techniques has largely focused on genetics. Advances in next-generation sequencing have made it relatively straightforward to screen solid tumor cancers for particular genetic variants. For example, individuals with HER2 mutation positive breast cancer can be treated with drugs that target the HER2 protein.¹

But for many cancer patients—particularly those with hematological cancers—no established genetic marker exists. To combat this deficiency, oncologists apply broader therapies, often with varied outcomes. In acute myeloid leukemia (AML), for example, a number of patients successfully treated with chemotherapy see their cancer return, and it’s not yet fully understood why.²

To help answer such questions, researchers are now looking beyond genetics to the use of proteomics, lipidomics and metabolomics in precision medicine.

These approaches rely on up or down regulation of proteins, cellular lipids or metabolites, which tend to vary over time and can be influenced by disease state. By studying the markers and how they track disease progression or treatment response, researchers could make precision medicine that much more precise.

For those working in biomarker discovery, one aim, ultimately, is to establish correlations between a biomarker and the likelihood that the patient will respond well to a specific treatment. But applications could be more nuanced still.

“We want to take that even further,” says Mike Baratta, a Senior Director at Takeda Pharmaceuticals. “If you have expression of a specific target over a cutpoint, we want to be able to predict what clinical response would be.”

First, clinicians need to know which biomarkers, or combinations of biomarkers, they should monitor, and implement robust methods and clinically validate them.

The search for biomarkers

While all the omics have a role to play in precision medicine, genomics remains the most mature approach in clinical applications. Proteomics strategies are developing rapidly, but their clinical potential will rely heavily on innovations in multiplexed mass spectrometry, which is capable of profiling thousands of proteins in a sample.

Researchers have, of course, been profiling proteins in blood for decades. But they mostly used enzyme-linked immunosorbent assays (ELISA), which use antibodies to capture and detect known proteins from a sample.³ Mass spectrometry does not always require researchers to know which proteins they’re seeking in advance, making it much more suitable for, say, biomarker discovery and characterization. It's not limited to proteins either; similar methods also can apply to lipidomics and metabolomics.

“We can use a completely unbiased [biomarker discovery] strategy,” says Paul Beresford, Chief Business Officer of the diagnostics developer Biodesix. “We can look very broadly at patient subsets that either benefit or not from a drug, and at pretty much all the proteins that we can see.”

Biodesix uses several distinct, mass spectrometry-based approaches for biomarker discovery. One pairs mass spectrometry with liquid chromatography (LCMS). This method was originally created for small molecules, but can also detect proteins when they're digested into peptides.⁴ It is a very sensitive technique that can discriminate the levels of key proteins in a blood sample. Certain levels of a protein, not just its presence, can indicate cancer diagnosis or its response to therapy.

Newer mass spectrometry methods further expand sensitivity. For example, the timsTOF method, developed by instrument company Bruker, combines trapped ion mobility spectrometry (TIMS) with time-of-flight (TOF) spectrometry to separate compounds even further, and identify peptides and proteins that would have escaped detection otherwise.⁵

Biodesix also uses a proprietary method called DeepMALDI analysis to identify patients that can benefit from therapies.⁶ This method is based on matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry, which can detect intact proteins.^7,8 “We’re currently at a point where we can measure a little over 2,200 proteins in serum and plasma and their post-translational modifications to look for patterns that are associated with a disease,” says Robert Georgantas, Senior Vice President of Research and Translational Science at Biodesix.

Regardless of the method used to discover potential biomarker candidates, Georgantas’ team uses either conventional statistics or an artificial intelligence method to further narrow down the pool of clinically significant candidates that might be suitable diagnostic tools.

Diagnose therapeutic mechanism, not just disease

Proteomics will almost certainly open avenues for precision medicine previously off-limits for hematological cancers. Beyond patient-therapy matching, it could help researchers understand how a treatment will progress over time. Unlike genetic profiles, protein expression changes over the course of disease progression or response to treatment. Proteomic biomarkers could help predict how well patients respond to treatment based on disease progression.

Such insight could be gathered from a single biomarker but, as Baratta has discovered, combinations of biomarkers and other variables might be more telling. In Takeda’s search for new biomarkers for hematological cancers, researchers found different biomarkers linked to remission and to time to follow-up. “You’d think that they would correlate,” Baratta says. Both are related to how much time has passed since initial treatment, but that’s not actually the case. Rather, Baratta says, “I think you need to cast that initial net fairly wide.”

That net could include other molecular biomarkers, or even age, gender, and whether or not this was a first cancer diagnosis. In the fall of 2019, Biodesix launched just such a test, the Nodify XL2 test, which measures two proteins along with 5 clinical factors. It is commercially available to physicians and covered by Medicare.⁹

While most of the focus today is on biomarker discovery, the ultimate goal is to validate biomarker use in diagnostics. That involves clinical trials, as well as considering scenarios with small sample volumes or other challenges that could come up in a clinical setting. While the path is relatively clear for such work, as Biodesix has shown with its approved diagnostics, success is never certain, fast nor easy.

But given time, it will come, says Beresford: “In personalised healthcare, you're going to see a wave of new technologies continue to improve how we identify the right patient for the right drug at the right time.” Georgantas agrees, adding that “therapeutic guidance is really going to be the next generation of where biomarkers are going.”