Wider access to liquid biopsy might be right around the corner.

Greg Goth

March 1, 2024

6 Min Read
Image Credit: Md Saiful Islam Khan via iStock/Getty Images

At a Glance

  • Both studies utilized machine learning to streamline the identification of cancer markers.
  • The distillation of relevant data through multi-omics analysis and machine learning holds promise for early cancer detection.
  • The studies shed light on the role of proteomics and future directions.

Were a well-meaning party guest to pull aside Dustin Hoffman’s Ben Braddock character in “The Graduate” today, the “just one word” he mentions as a key to Ben’s future might very well not be “plastics,” but rather “multi-omics.”

This is especially the case when it comes to emerging liquid biopsy research.

Just days after The Association for Diagnostics & Laboratory Medicine published a special edition of its Clinical Chemistry journal highlighting the field, two studies highlighting multi-omic breakthroughs in liquid biopsy emerged. One, conducted by San Mateo, Calif.-based PrognomiQ and published in medRxiv, looked at more than 2,500 subjects at risk for lung cancer. The other, led by researchers at Cedars-Sinai Medical Center in Los Angeles and published in Nature Cancer, found new biomarkers for pancreatic cancer through multi-omics and the medical center’s Molecular Twin precision oncology platform. The study analyzed blood and tissue samples from 74 patients with the most common and most aggressive pancreatic cancer type, pancreatic ductal adenocarcinoma. The researchers found their method and technology outperformed the only Food and Drug Administration-approved pancreatic cancer test, a blood test called CA 19-9. The findings were validated in independent datasets from The Cancer Genome Atlas, Massachusetts General Hospital, and Johns Hopkins University.

Both studies also used machine learning to significantly streamline the markers that most convincingly signaled cancer might be present. The PrognomiQ study’s profiling detected 113,671 peptides corresponding to 8385 protein groups, 219,729 RNA transcripts, 71,756 RNA introns, and 1801 metabolites across all subject samples; they then developed a machine learning-based classifier comprising 682 of these multi-omics analytes. The Cedars-Sinai team first combined 6,363 different biological data points, including genetic and molecular information, to create a model that accurately predicted disease survival in 87% of patients. The team then used AI to streamline the data and create a model that performed nearly as well with just 589 points of data.

Such distillation of relevant data may be a key to making liquid biopsy a clinically viable technology that can either complement or, perhaps, eventually replace traditional tissue biopsies in detecting early cancers or better classify their malignancy severity. They show potential to significantly reduce barriers to the detection of early-stage cancers and to better classify those that are found.

For example, in announcing the results of its study, PrognomiQ stated that the current standard of care for detecting early-stage lung cancer in the United States, low-dose computed tomography, stands at a disappointing 10% adherence for high-risk individuals, 10 years after that standard was announced.

After modeling their data in their ML algorithm, the PrognomiQ researchers said they found features representing 149 distinct proteins, 346 distinct genes, and 77 distinct metabolic pathways that facilitate further development of a practicable assay for early detection of lung cancer.

“Clinical development of this assay would address a critical clinical need for early cancer detection given the classifier’s performance at detecting stage I disease, which comprised a majority (51.8%) of lung cancer cases in the National Lung Screening Trial,” they wrote.

Likewise, the Cedars-Sinai team’s streamlining of the original 6,363 data features that accurately predicted pancreatic cancer survival rates down to 589 might also portend the ability to extend highly accurate and cost-effective liquid biopsy to people with no such access now.

“Our platform enables discovery of parsimonious biomarker panels and performance assessment of outcome prediction models learning from resource-intensive panels,” they wrote. “This approach has considerable potential to impact clinical care and democratize precision cancer medicine worldwide.”

Both studies also cited recent advances in protein analysis, or proteomics, as enabling their breakthroughs. PrognomiQ CEO and founder Philip Ma said the new capabilities are bringing proteomics along with genomics to pinpoint cancerous anomalies; the company is a spinout of Seer, which developed the Proteograph proteomics suite.

“The thing about the gene is that there is a mechanism to amplify it because genes replicate – technologies that enable sequencing to amplify the gene,” Ma said. “Protein doesn’t have that, so you have to have quite complex workflows to do that. It would involve literally thousands of hours.”

Likewise, the Cedars-Sinai team cited proteomics as key to their breakthrough.

“Once a patient has cancer, proteins act as the body’s first responders, and their activity helps us determine how a patient’s body is reacting,” co-author Jennifer Van Eyk said in the announcement of the results. “Proteins turned out to be the main drivers of our pancreatic cancer models. And in future studies, proteins will also help us track how well a patient is responding to treatment.”

No AI needed for prostate biopsy discovery

One possible breakthrough in liquid biopsy research that did not use AI was undertaken by another team at Cedars-Sinai investigating improved prostate cancer detection; their findings were published in nanotoday. The nano-technology used what the study’s corresponding author, Edwin Posadas, M.D.,  medical director of the urologic oncology program and co-director of the experimental therapeutics program at Cedars-Sinai Cancer, said was a “simple” polymerase chain reaction (PCR) technique to analyze extracellular vesicles (EV’s) derived exclusively from prostate cancer.

As Posadas and his colleagues explained, EVs are phospholipid bilayer-enclosed particles released from both normal and cancer cells into body fluids such as urine and blood. By transferring biomolecules between cells, EVs can function as a vehicle for intercellular communication. This process has also been shown to prime the pre-metastatic niche at distant organs and facilitate the invasion and dissemination of cancer cells.

“We’ve been building this assay and incorporating it into several studies that are ongoing,” Posadas said. “One studies men whose PSA’s return after surgery. We believe we may be able to see a signature of a greater degree of aggressiveness that could help us decide that some men would benefit from hormonal therapy in addition to salvage radiation.”

Posadas said such precision in novel technologies such as the one his team is perfecting will be an important adjunct to ultra-sensitive technologies already in common clinical use in monitoring both pre- and post-treatment cancer presence. For instance, new highly sensitive PSA tests that return results in the hundredths of nanograms per milliliter can either give a patient false reassurance or cause unnecessary worry. And, while the traditional PSA reading signaling the biochemical return of cancer is 0.2, a sub-clinical reading well below that in two different patients can mean two different things.

“Your disease at .08 can have a pretty benign engine to it, not taking you to a place where you have metastasized, but another guy could have a PSA of .05, and we see something that might signal you would want to do something before you get to 0.2,” Posadas said. “By the time you get there, there may be more cancer in you than in the guy sitting next to you with the same PSA, because you have a different program to your cancer. This is the problem I wrestle with in the clinic whenever I have to tell a guy what his PSA means in the big picture.”

Liquid biopsy may not emerge immediately as ubiquitous front-line diagnostic technology; both Cedars-Sinai studies were done on small numbers of patients, and PrognomiQ is set to test its technology on a trial Ma said could enroll up to 15,000 patients. Yet Posadas said the success of these studies indicates a bright future for the field.

“The technologies are out there, but we haven’t shown a big enough benefit for the doctors in the community to say they need to invest in getting access to this resource because it helps patients,” he said. “But I would probably tell you that in the next five years, you’ll see more of these liquid biopsy tests being deployed.”

About the Author(s)

Greg Goth

Greg Goth is a freelance technology writer.

Sign up for the QMED & MD+DI Daily newsletter.

You May Also Like